Spike is a B2B Generative AI and Data tech startup for the Digital Health industry. It provides health and fitness data interpretation-as-a-service on top of user-generated data from Wearables and IoT devices. Spike provides 3 Core products: 1) API aggregation (access to 200+ sensors via a single API) 2) Data pipeline (managed-ETL tool allows to get data in low-code and fast) 3) GAI-based Health Data Interpretation SDK (plug&play AI for text-based health observation or interpretation for their end users) Data from wearable sensors, smartwatches, IoT devices are actively used throughout the healthcare industry. That said, only a portion of the industry knows how to leverage the data to its potential. Using Spike, you can integrate this data from commercially used optical, electrical, electrochemical, and electromechanical sensor types. These devices extract biometric data such as heart rate variability, calories, blood pressure, glucose and cortisol levels, and others. Spike enables you to import these metrics directly into your cloud via a managed ETL service while providing AI tools for recommendation and prevention.

Spike

spike

Move over, Dr. House. There's a new medical genius in town, and it doesn't even have a stethoscope. Spike, a Silicon Valley startup that focuses on data technology and artificial intelligence (AI), just landed $3.5 million in seed funding. Its mission? To give healthcare AI, which is set to reach $148 billion by 2029, the brainpower it needs. The oversubscribed investment was co-led by Practica Capital and TheVentureCity, and a who's who of tech investors joined the party, including CEAS Investments, Plug & Play Ventures, Geek Ventures, and Axel Springer Porsche (APX). Spike is ready to supercharge product development, expand into new markets, and cement its status as a medical director in the AI health technology field. So, what's Spike's secret remedy? Its B2B solution allows developers to add AI superpowers to their healthcare apps. Spike aims to cut development time and costs to a fraction of what they used to be and connect to the data collected by hundreds of medical, wearable, and IoT devices. Think of Spike's API as a health data buffet, offering insights from over 300 sources – from your trusty Fitbit to high-tech medical devices. It's the fuel that powers Spike's AI algorithms to deliver personalized healthcare solutions smarter than a med school grad (well, almost). "Advancements in retrieval-augmented generation AI mean that LLMs are no longer taking a 'best guess,' but are now able to seek out specialized context, producing outputs that are much more accurate and considerably less prone to hallucinations," Povilas Gudzius, CEO and co-founder of Spike, said. "By incorporating this technology on top of Spike's high-quality health data API, we're enabling our clients to easily leverage these tools and improve the health of millions of end users." Spike vs. the Goliaths, with a RAG, not a Stone Sure, Google, Amazon, and Microsoft have their AI health data solutions. But let's be honest: They're like old-school doctors who still use pagers. Spike is the young, energetic upstart shaking things up. Google Cloud Healthcare API provides access to healthcare data and offers AI and machine learning tools for building healthcare applications. Amazon HealthLake is a HIPAA-eligible service that helps healthcare organizations store, transform, query, and analyze health data. And Microsoft Azure API for FHIR enables the exchange of healthcare data using the FHIR standard and offers AI tools for building healthcare applications. So while the megaliths in the room are busy managing massive cloud infrastructures, Spike is laser-focused on healthcare. They're not just aggregating data; they're curating it, ensuring it's top-notch and tailor-made for the unique needs of the medical world. And Spike uses retrieval-augmented generation (RAG), a technique for enhancing the accuracy and reliability of generative AI models with facts fetched from external sources. RAG helps AI models avoid embarrassing "hallucination" moments by cross-checking their answers with real-world facts. Think of it as a reality check for AI, ensuring its answers are as accurate as a doctor's diagnosis. This makes AI more reliable, less prone to making things up, and ultimately, much more useful in the healthcare world. "We are utilizing health expertise networks and curated in specialized context databases in RAG AI infrastructure," Gudzius said. "RAG AI allows us to incorporate these problem-specific sources effectively and makes the LLM responses more grounded. The design of the RAG AI infrastructure and domain-expert curated medical content allows us to further innovate in developing the 'Trustworthy Health AI.'" All of this technology doesn't mean Spike can scrimp on security and privacy. Like the Big 3, it has to make strong commitments to regulatory protection, especially with searches for AI in healthcare being orders of magnitude higher than searches for AI safety. "Spike takes data security and privacy very seriously, so our infrastructure products follow strict GDRP, HIPAA, and CCPA regulatory guidelines," Gudzius said. "From the technical perspective, we are also adopting specific techniques to prioritize data privacy. For example, we have built our data access layers that provide limited control to the LLM agent while handling data acquisition logic and complying with regulatory requirements. As the quality and volume of individuals' health data grows, there's a huge need to develop and innovate regulatory-compliant techniques for private health data-to-LLM interactions like that." The Future of Healthcare: Spike's Prescription This $3.5 million isn't just pocket change; it's the catalyst for Spike's grand vision. They're not just building AI tools but trustworthy AI tools that are as reliable as your favorite family doctor (but hopefully with a better bedside manner). They're expanding their reach to include everything from electronic health records to genomics data. "We believe in a future where health data volume will grow exponentially driven by the improvements in sensor hardware," Gudzius said. "These sensors include optical, electrical, and electromechanical sensors from wearables, IoT devices, and other sources like EHRs and lab tests. We are well positioned to capture this potential because our team has extensive experience working with data acquisition and a scientific background in Health AI. Our edge is ensuring we provide the capabilities of these emerging non-diagnostic technologies to our clients in a safe and regulatory compliant way." Spike imagines a future where AI isn't just a buzzword; it's an integral part of healthcare, improving patient care and making those dreaded doctor's visits a little less dreadful. A version of this article was originally published on Dataconomy and is republished with permission. Move over, Dr. House. There's a new medical genius in town, and it doesn't even have a stethoscope. Spike , a Silicon Valley startup that focuses on data technology and artificial intelligence (AI), just landed $3.5 million in seed funding. Its mission? To give healthcare AI, which is set to reach $148 billion by 2029 , the brainpower it needs. Spike set to reach $148 billion by 2029 The oversubscribed investment was co-led by Practica Capital and TheVentureCity, and a who's who of tech investors joined the party, including CEAS Investments, Plug & Play Ventures, Geek Ventures, and Axel Springer Porsche (APX). Spike is ready to supercharge product development, expand into new markets, and cement its status as a medical director in the AI health technology field. So, what's Spike's secret remedy? Its B2B solution allows developers to add AI superpowers to their healthcare apps. Spike aims to cut development time and costs to a fraction of what they used to be and connect to the data collected by hundreds of medical, wearable, and IoT devices. Think of Spike's API as a health data buffet, offering insights from over 300 sources – from your trusty Fitbit to high-tech medical devices. It's the fuel that powers Spike's AI algorithms to deliver personalized healthcare solutions smarter than a med school grad (well, almost). "Advancements in retrieval-augmented generation AI mean that LLMs are no longer taking a 'best guess,' but are now able to seek out specialized context, producing outputs that are much more accurate and considerably less prone to hallucinations," Povilas Gudzius, CEO and co-founder of Spike, said. "By incorporating this technology on top of Spike's high-quality health data API, we're enabling our clients to easily leverage these tools and improve the health of millions of end users." Spike vs. the Goliaths, with a RAG, not a Stone Spike vs. the Goliaths, with a RAG, not a Stone Sure, Google, Amazon, and Microsoft have their AI health data solutions. But let's be honest: They're like old-school doctors who still use pagers. Spike is the young, energetic upstart shaking things up. Google Cloud Healthcare API provides access to healthcare data and offers AI and machine learning tools for building healthcare applications. Amazon HealthLake is a HIPAA-eligible service that helps healthcare organizations store, transform, query, and analyze health data. And Microsoft Azure API for FHIR enables the exchange of healthcare data using the FHIR standard and offers AI tools for building healthcare applications. So while the megaliths in the room are busy managing massive cloud infrastructures, Spike is laser-focused on healthcare. They're not just aggregating data; they're curating it, ensuring it's top-notch and tailor-made for the unique needs of the medical world. And Spike uses retrieval-augmented generation (RAG), a technique for enhancing the accuracy and reliability of generative AI models with facts fetched from external sources. RAG helps AI models avoid embarrassing "hallucination" moments by cross-checking their answers with real-world facts. Think of it as a reality check for AI, ensuring its answers are as accurate as a doctor's diagnosis. This makes AI more reliable, less prone to making things up, and ultimately, much more useful in the healthcare world. "We are utilizing health expertise networks and curated in specialized context databases in RAG AI infrastructure," Gudzius said. "RAG AI allows us to incorporate these problem-specific sources effectively and makes the LLM responses more grounded. The design of the RAG AI infrastructure and domain-expert curated medical content allows us to further innovate in developing the 'Trustworthy Health AI.'" All of this technology doesn't mean Spike can scrimp on security and privacy. Like the Big 3, it has to make strong commitments to regulatory protection, especially with searches for AI in healthcare being orders of magnitude higher than searches for AI safety . searches for AI in healthcare being orders of magnitude higher than searches for AI safety "Spike takes data security and privacy very seriously, so our infrastructure products follow strict GDRP, HIPAA, and CCPA regulatory guidelines," Gudzius said. "From the technical perspective, we are also adopting specific techniques to prioritize data privacy. For example, we have built our data access layers that provide limited control to the LLM agent while handling data acquisition logic and complying with regulatory requirements. As the quality and volume of individuals' health data grows, there's a huge need to develop and innovate regulatory-compliant techniques for private health data-to-LLM interactions like that." The Future of Healthcare: Spike's Prescription The Future of Healthcare: Spike's Prescription This $3.5 million isn't just pocket change; it's the catalyst for Spike's grand vision. They're not just building AI tools but trustworthy AI tools that are as reliable as your favorite family doctor (but hopefully with a better bedside manner). They're expanding their reach to include everything from electronic health records to genomics data. "We believe in a future where health data volume will grow exponentially driven by the improvements in sensor hardware," Gudzius said. "These sensors include optical, electrical, and electromechanical sensors from wearables, IoT devices, and other sources like EHRs and lab tests. We are well positioned to capture this potential because our team has extensive experience working with data acquisition and a scientific background in Health AI. Our edge is ensuring we provide the capabilities of these emerging non-diagnostic technologies to our clients in a safe and regulatory compliant way." Spike imagines a future where AI isn't just a buzzword; it's an integral part of healthcare, improving patient care and making those dreaded doctor's visits a little less dreadful. A version of this article was originally published on Dataconomy and is republished with permission. A version of this article was originally published on Dataconomy and is republished with permission. published on Dataconomy

Spike Aims to Give Healthcare AI a Brain Boost and Outsmart the Tech Giants

Welcome to my meet the entrepreneur Interview Series, where I interview impactful leaders and ask them to share tips and techniques that have benefited them, both in their personal and professional lives. This interview is with , Co-Founder & CEO of . Dvir Ben Aroya Spike About Dvir: Dvir has over 20 years of executive experience leading technology and internet companies. As an entrepreneur, he is passionate about creating SaaS businesses and nurturing them throughout their entire lifecycle. When he's not talking about how to improve modern communication, you can find him surfing with friends or running after his three boys. How Do You Deal With Stress And Anxiety? It is important to know that within any industry, there are good and bad days–with that being said, I always try to allocate time to things that help me relieve stress. For me, this is through sports. Wave surfing is my form of relaxation that makes me feel centered. Along with sports, another thing that helps me destress is spending time with family; it's good to take a break and keep things in proportion. Your Tips For Aspiring Entrepreneurs? No matter where you are in your business journey, it will always feel like you have a million tasks to tackle on a daily basis. One tip would be to always prioritize tasks that will have the maximum impact on your business. Things will come your way fast, and you will need to know how to use your time wisely in order to meet your end goals. Another tip is to surround yourself with like-minded and knowledgeable entrepreneurs who constantly push you to learn, grow and think differently. Who Would You Like To Meet For Lunch and Why? Salesforce Founder and CEO Marc Russell Benioff. Salesforce, a leading CRM cloud-based software company, disrupted the market and created the software SaaS vertical. Benioff highlights how his experiences and past history are a big factor in his successes today, and I would find it interesting to sit and learn how his past experiences led him to shape Salesforce. What Is Your Favorite Book Or Movie Recommendation? The High Growth Handbook by Elad Gil: . This book gives good advice and key lessons for startups and entrepreneurs. https://growth.eladgil.com/book/introduction/welcome-to-the-high-growth-handbook/ What Habit Has Contributed To Your Career? Perseverance. Things don't come easily; sometimes, you need to fail 100 times in order to succeed once. I make it a habit to show up and keep trying no matter what the outcome. When You Hear About Other Famous Entrepreneurs, Is There Any Entrepreneur Who Specifically Caught Your Attention? Throughout my career, I have met many entrepreneurs and interesting people who have taught me new and innovative things. If I had to choose one who specifically caught my attention, I'd say the CEO of Zoom, Eric S. Yuan. He was able to take a technology that has been around for years and turn it into something extremely successful based on feedback from his customers and audience to ultimately connect the world. Is There a Specific Business Philosophy That Has Helped You? To constantly listen to feedback– from customers, team members, and your community. Things are versatile and ever-changing; it is important to constantly evolve and accept change to align with growth. About Spike: is the world's first conversational email app. They are upgrading the way SMBs work. Spike's goal is simple: bring all communication -- emails, chats, calls, team collaborations, tasks, everything -- to one place. It's what teams have been looking for in productivity and messaging apps, but easier and all from your inbox. Spike Spike works on top of any existing email (O365, G suite, and IMAP) and is available on iOS, Android, Mac, Windows & Web. Also published . here

Meet The Entrepreneur: Dvir Ben Aroya, Co-Founder & CEO, Spike 

During refinement a team member says: . She warns we should not start working on the feature before doing a spike. ”I have no clue how to build this feature” Mentioning spikes often creates confusion in a team. Every team member has their own opinion about what a spike is and the right time to use one. I was part of teams that wanted a spike for almost everything they worked on. I also worked with teams that rarely used spikes. Both approaches were suboptimal. When is using a spike appropriate? Before we expand on when and how spikes should be used, let’s talk about the origin of the term ‘Spike’. The term spike stems from Extreme Programming (XP) and has it’s roots in rock climbing. When rock climbing, you sometimes cannot climb further because there is nothing to grab on to. To climb further, you need to drive a spike into the rock surface. The spike does not move you closer to the top, but . forges a path for you to continue climbing on A spike in software development serves a similar purpose. Spikes help to move forward when . knowledge is lacking to build something When should you use a spike? Backlog refinement is all about removing uncertainty and creating common understanding. Lack of knowledge is the starting point in refinement. So, it is important to understand when using a spike is appropriate or overkill. In product three types of uncertainty are discussed with the development team: backlog refinement Why do we need to build this? Purpose uncertainty. What do we need to build? End uncertainty. How can we build it? Means uncertainty. At the start of refinement, everything is unclear. After refining a feature, all three types of uncertainty should be sufficiently clear to be able to estimate the feature. If your team does not use estimates, you should be able to split the feature up in sprint-sized chunks. Sometimes everything is clear, except how to build a feature. If the team has no idea how to build the feature, then the work cannot be accurately estimated. It cannot be split up to fit in a single sprint either. The risk of building a technical solution that is unfeasible increases. Now is the right moment to use a spike. When the team has no idea to build a feature, then planning a is appropriate. how spike A spike is an investigation the team has to perform they can start working on the feature it relates to. The outcome of the spike needs to be explicitly defined up-front. before A spike is usually time-boxed. This means the investigation is over when the time-box expires, regardless whether the spike outcome is met or not. A time-box helps to manage time and prevent endless research without results. By definition, research is more open-ended and divergent than delivering something. When a spike is completed (with or without the time-box expiring), there are two options: You have enough information on the technical solution to proceed. You can now estimate the feature. If your team does not use estimates, you have enough information to split the feature up in chunks that take less than a sprint. The investigation fails to provide enough information on the technical solution to proceed. Work cannot be started yet, unless the risk of being unable to finish the feature in a single sprint is accepted. Sometimes people make the distinction between functional and technical spikes. The only difference is the type of research. A is planned when it is unclear how to build something. A is planned when more information is needed to decide what to build from functional perspective. technical spike functional spike Now that you have basic understanding about what a spike is and when you should use one, let’s discuss 6 common mistakes teams make when using spikes. 1. Performing a spike when many other things are unclear If the purpose of the backlog item or how it should work are unclear, then you are better off reducing these types of uncertainty first. Otherwise, how can you be sure that a spike is necessary? With a better understanding of the problem, you give the team more freedom to come up with a suitable technical solution. Uncertainty in the ‘why’ and ‘what’ propagates to the technical solution. Maybe there is a different way of solving the problem your team does not consider, because they do not understand the problem well enough. 2. Using spikes for implementation details you can figure out as you go along A spike should give your team confidence it can build a feature and split it up in chunks that fit in a sprint. It should not be used to figure out every little detail before you can work on a feature. Some teams will use spikes to nail down all details. They believe that everything needs to be crystal clear before you can work on a feature. The team wastes time researching implementation details that could be figured out as you go along. As a result, sprints alternate between research and delivery work. The time-to-market of features increases. Features requiring a spike will take at least two sprints to deliver, because you need to do the spike one sprint before. If the team is confident it can finish the spike and feature it belongs to in a single sprint, why do the spike at all? Figuring out implementation details during the same sprint you perform the work is more effective. You minimize waste, hand-overs and doing up-front analysis. The feature will also be released sooner, allowing you to process valuable market feedback earlier. 3. Doing spikes before considering to reduce scope When a team member says a spike is necessary, you should try to understand what technical part makes the solution difficult. If possible, just remove the difficult part from the functional scope. Just pick up what is easy first and worry about the hard part later. This way the feature can be estimated and picked up without any preceding spike. Sometimes just by starting to work on a feature the team acquires enough knowledge to work on the difficult part. If this is not the case, then at least you will have more knowledge before doing the spike. 4. Not defining a clear spike outcome Before working on a spike, the expected outcome of the investigation should be clear. By making the expected outcome clear, you prevent your team from being unable to move forward even though the spike has been completed. What is the information you need to research in order to move forward? Should it be a working proof-of-concept? Or is a document that describes the technical solution good enough? What will give the team enough confidence to move forward? 5. Executing a spike long before you need it Analysis work many sprints before you start working on a feature is waste you should try to prevent. When an spike has been completed, the outcome is fresh. As time passes on, spikes become stale just like food. Stale food is not enjoyable and neither are stale spikes! The system landscape may have changed so much, the spike is no longer relevant. It might also be that the feature is no longer important, because it no longer fits the direction of the product. As time passes on, the team loses knowledge it gained through the spike. The team will need more time to understand the outcome of the original investigation. You should try to plan a spike as close to the moment you want to pick-up the backlog item to minimize all these different kinds of waste. 6. Planning spikes without digging deeper When your team keeps asking to plan spikes, this could point to bigger problems the team is facing, such as: Knowledge the team misses in specific areas or a single team member having most of the knowledge. Legacy systems that are very difficult to maintain or adjust. Nobody feels comfortable estimating the work. If this keeps on happening, it may pay off to get to the bottom of these problems . To see if there is something you can do to prevent these spikes from being necessary. Maybe it pays off to refactor some parts of the system instead of performing the spike. It could also be you just need to plan a knowledge sharing session, so the team gains knowledge about part(s) of the system or a specific framework. Spikes help manage uncertainty, but should be used sparingly Spikes should be exceptions, not the rule. You should ask these questions before deciding if a spike is appropriate: Does the team understand the problem well enough to come up with the best solution? Get everything else that is uncertain, as clear as possible, before you decide a spike is necessary. Uncertainty in why we are building something and what we need to build propagates to the technical solution. Is the spike really necessary? Sometimes you can figure out the technical details as you go along. Some uncertainty is to be expected. Is the team suffering from a waterfall mindset or is the spike justified? Can the scope be reduced to move what is difficult out of scope? This way you can start working on the feature without a spike. After finishing the feature, sometimes you will have gained enough knowledge to work on the difficult part without a spike. If you still need one, then at least you will have more knowledge before starting the investigation. Did the team define a clear expected outcome in the spike? The expected outcome of the spike needs to be clear to minimize the chance of failure. Will the feature where the spike relates to be picked up soon? Spikes become stale. Changes in technical landscape or business priority may make spikes obsolete. So plan spikes as close to the actual work as possible to prevent waste. Is the excessive usage of spikes not symptomatic of bigger problems? Dig deeper to find out what problems your team is facing. Then make a decision if it is better to use a spike or spend time to address the root cause.

6 common mistakes when using spikes

**GRAND CAYMAN, Cayman Islands, March 28th, 2024/Chainwire/--**March DEX volume on Sui stands at over $2.88B – up more than 49% from February – with decentralized exchange Cetus and wholesale liquidity layer DeepBook leading. Sui, the fast ascending Layer 1 blockchain offering industry-leading performance and infinite scaling, continued to demonstrate its blistering DeFi growth by cementing its place in the top ten of all chains by weekly DeFi volume, registering over $830M of trading over the most recent seven days of data available on and $224M in the latest 24 hours. DeFiLlama The volume growth comes as the chain has amassed several DeFi milestones since the start of the year, including surging past $700M in total value locked (TVL) earlier this week – up over 1900% since October – and dominating outflows from Ethereum via Wormhole during multiple seven-day periods this year. Sui is also setting the standard for network reliability during periods of high network traffic. In the first four months of its mainnet, Sui completed 65.8M transactions in a single day—the most of any blockchain ever, while throughout its existence, the network has never had an outage or experienced degraded performance. These milestones reflect how Sui’s underlying technology, which features an object-centric model, horizontal scalability, the increasingly popular, developer-friendly Move programming language, and recent breakthroughs such as zkLogin and zkSend, enables developers to create products that solve real-world challenges at scale. Sui’s technology has also attracted top projects and developers which are increasingly choosing to build on Sui. Most recently, , a protocol developed by the team behind Solana’s largest lending protocol, Solend, launched in its first expansion outside the Solana ecosystem. Ondo Finance’s treasury-backed assets also recently made their way into the ecosystem, while decentralized derivatives exchange left Arbitrum for Sui. Suilend Bluefin DeepBook, Sui’s fully on-chain order book, which is part of the Sui Protocol, has experienced a substantial volume increase in March as DeFi on the network continues to flourish. The CLOB saw over $940M in volume since the start of the month, and over $289M of trading in the last week alone, as it continues to provide deep liquidity to all DeFi protocols and traders on Sui with ultra-low, predictable transaction costs and sub-second finality. “From the Move language to its object-centric architecture, Sui’s technology is unique in the industry and enables a level of innovation that is attracting partnerships, protocol expansions, and independent developers—all building solutions that leverage the Sui network,” said Greg Siourounis, Managing Director of the Sui Foundation. “Sui’s surging metrics in DeFi indicate definitively that in an industry which is currently surging and primed for cutting-edge technology and practical, real-world applications, Sui is increasingly the platform of choice.” Contact Sui Foundation media@sui.io This story was distributed as a release by Chainwire under HackerNoon’s Business Blogging Program. Learn more about the program here.

Sui Spikes In Weekly DEX Volume, Joins Top 10 Of All Blockchains

Donald Trump, the former president of the U.S., saw an increased interest in his NFT collection after being arrested and charged at a Manhattan court. Reports suggest that Trump Digital Trading Cards sold 47 tokens within 24 hours of his arrest marking, an increase of over 400%. Contents Trump NFT Sales Rise The trading volume for the billionaire’s Trump Digital Cards collection has increased astronomically by more than 220%. Reports from some news outlets maintain that sales of $75,000 worth of Trump NFT were recorded within a day! As the number of sales of transactions went up, the collection’s price witnessed a similar occurrence.  Over 60 consumers bought the tokens within half a day, and a 4.3% increase left its cheapest NFT priced at 0.47 ETH, which is currently the equivalent of $908. Ironically, Trump used to be a loud disparager of crypto, most significantly, BTC. I am not a fan of Bitcoin and other cryptocurrencies, which are not money, and whose value is highly volatile and based on thin air, he famously tweeted back in 2019. https://twitter.com/realDonaldTrump/status/1149472282584072192?s=20&embedable=true Barely three years later, in 2022, he launched his personal NFT collection. His collection has been so successful lately that someone who went by “paraguanads.eth” purchased a card with Trump holding a rifle, draped in an orange hat for 0.687 ETH (the equivalent of $1,333). In addition, Trump’s accomplishment further boosted the overall NFT market capitalization by 7% to 21,150 ETH. The Trump Digital Cards, minted on Ethereum and Polygon, and released in 2022 for $99 each, didn’t take long to get sold out. They also came bearing gifts, as buying 45 NFTs enables customers to meet him for golf or dinner at the Mar-a-Lago club he owns. However, how did all of these come about? Why was Trump arrested? Donald Trump, who is the first ex-American president to be charged to court, was criminally indicted in New York over 34 felony cases for allegedly forging business records. He argued his innocence and pled not guilty to the allegations. He allegedly fabricated business records over hush-money payments involving him and an ex-porn actress, Stephanie Clifford, also called Stormy Daniels. Trump was said to have been in an extramarital affair with her, who asserted that they were once in a sexual relationship as far back as 2016. This occurred during the 2016 presidential campaign, and knowing fully well that this could hurt his ambitions, he decided to settle her quickly and quietly. His lawyer, Michael Cohen, paid Daniels $130,000 for her to sign a non-disclosure agreement (NDA) that prevented her from publicly speaking about the matter. Incidentally, Cohen pleaded guilty to an excessive campaign charge in 2018. He further disclosed that he made this payment on behalf of Trump, who was aware of and ratified it. Trump’s NFT collection enjoyed a historic and astronomical rise, but it remains to be seen what comes of it. Also published . here

Exploring the Spike in Trump's NFT Sales Following His Arrest

While Slack has been dominating the domain of enterprise communications for a long time, more innovative alternatives have emerged that revolutionize how we maintain communications in the workplace and not only. takes e-mail into the current digital reality and equips your inbox with every productivity tool you need; it is the ideal hub for corporate communications. Spike With Spike, all the information you need to complete your task is included in a single stream, so you can finally stop hopping between applications. Spike's conversational e-mail combines conventional e-mail with instant chat, hence avoiding convoluted e-mail chains. With real-time chat and real-time collaboration, you will have superior interactions and a more human communication experience. Spike is available for iOS, Android, Mac OS X, Windows, and the Web. It is ideal for anybody seeking an effective and convenient method of communication because of its user-friendly design and various capabilities. Exploring Conversational Capabilities Slack is a messaging application that streamlines communication between teams. Well, the difficulty is that Slack is designed for internal teams, making it challenging to communicate with other parties, such as new customers and enterprises. In addition, it overcrowds communication channels and adds yet more application to the numerous already present. This might induce tension and disarray by increasing the amount of time spent switching between communication applications. Since its beginning, Spike has been committed to becoming the most effective corporate communications platform. So far, Spike has been embraced by some of the world's top organizations, including Fiverr, Zoom, and Shopify, and more than 150,000 teams and over 1.5 million users have enhanced their e-mail experience with Spike. Furthermore, Spike converts your current e-mail account into a streamlined conversation platform, while everyone else continues to get your e-mail in its original format. You may sort by contact as opposed to a topic, allowing you to see the whole thread history for any contact you have ever e-mailed! Rather than installing yet another chat program, it is incorporated within your e-mail. You may establish group conversations, collaborate on Notes, and synchronize your calendar across various e-mail accounts. Searching Features Overall, the free Slack account only allows users to see the 10,000 most recent messages. If you have a Pro account or above, you get access to the whole chat history and may search for any term or phrase inside the talks. This allows you to easily find what you are looking for, whether it is something for a team project or a specific document. Spike, unlike Slack, offers a Super Search that enables you to find any term, phrase, note, document, or file sent or received in the inbox. Even in group conversations, you can quickly and easily see all files in each thread. Additional Characteristics With Spike, there is a multitude of tools to explore, including unified calendars, team papers, and project management monitoring. Spike integrates them all into one platform accessible on your computer, mobile device, and online. Switching between a dozen distinct applications to complete a task costs time, energy, and resources. Spike, accessible wherever you are, enables a seamless workflow to keep team members on track and engaged. Moreover, Spike's built-in e-mail notetaking function makes it much simpler to write brief notes from inside the app without leaving the platform. Spike: The Future of E-mail & Digital Communications Well, Spike has you covered regardless of whether you need to communicate with friends, family, or colleagues. Therefore, give it a go and check out this breakthrough platform by accessing the following link: https://www.spikenow.com/

How Spike Replaces Slack as the Leading Communications Tool

“I’d get a message once a week that someone had used my picture. And then it just snowballed. They were stealing money and defrauding them using my image.” said Tom Ernsting, who is an Instagram model. This is catfishing in general and it’s been happening on major social media platforms. Catfishing Is Pathetic and Disrespectful Catfishing is a term used when a stranger pretends to be someone else and hides their social media information. For instance, boys who pretend to be girls online or the other way around, which is way more common. Now, strangers are using pictures and private information of another person to scam other people online, make money out of false relationships or tarnish a celebrity’s reputation. Catfishing was coined in 2010 and has become increasingly common since then as connected devices are more pervasive now. In February 18, 2021, the police arrested a fraudulent group in Manhattan, who was involved in catfishing and made . $55 million What is the Point of Catfishing? The simple answer can be: making money from false relationships, but we will drill down to the main catfishing reasons. If a person is lonely, depressed, and has no social connections in real life, he is more likely to get involved in catfishing to boost their self-esteem. Some people are trolls and use false identities to tarnish people’s reputations. Catfishing in this way can be out of pure jealousy. Taking revenge on a person you hate or despise can be another reason for catfishing. You can find a lot of catfishing cases that include an ex-boyfriend or ex-husband. People who don’t identify themselves as male or female might also create a false identity online to discover their sexuality. Examples of Catfishing 1. Falling in Love With a Scottish Model Who Turned Out to be Her Friend Laura Anderson, a Scottish star in a reality TV show Love Island, had a similar catfishing experience. She shared a story about a guy who thought he was in love with Laura after three months of exchanging text messages, but it turns out to be a stranger. She was shell-shocked that how easy it is to fool online users and create an Instagram story about the whole problem. 2. Catfishing My Best Friend for 10 Years Michonne shared her story about talking to a guy on Myspace who was lovely and obsessed with tattoos. However, the guy never showed up for a date or a hangout plan. That guy is also a friend of Michonne’s girlfriend, Amanda, who was actually catfishing Michonne for many years. 3. Faking It Big Time on TikTok Here is another catfishing problem that is way too common that we know. Individuals can create a profile and fake themselves as a celebrity or a self-help Guru to mint money from lonely individuals. The Effects of Catfishing Catfishing can take a toll on your mental health. Imagine a person who was emotionally attached to a stranger online, shared his inside feelings, and the stranger turns out to be an impostor. Catfishing can hurt your mental space, and the worst part, you will lose trust in humans. Further, there is also an element of embarrassment and financial loss. A scammer online can fool you easily if you are using dating apps and ask you to invest in crypto coins or real estate investments. All of this happens after weeks of back and forth communication between two people. Catfishing is scary, and there is no direct way to figure out if you are being catfished or not. Some Red Flags That You Are Being Catfished If your new online friend or crush is talking to you every day, here are a few signs you should be wary of and ensure you are not a victim of catfishing: Refusing to start a video call or face-to-face conversation Creating a sob story or shady back up story to start a relationship Sharing professional pictures on social media, Snapchat, or Instagram Avoiding a date night or an in-person meeting Expressing love and affection without getting to know you first How to Avoid Being Catfished With a wide array of online dating sites and apps available, catfishing a person is easier than before. People no longer mingle or hook up in a bar; they use dating apps and swipe right to meet new people. According to a survey, more than 30% of people online have been catfished. If you want to avoid being catfished, here are a few tricks you need to learn: Ask family and friends If you have started talking to a stranger online, the easiest way to find out if he or she is real is to ask your close loved ones or friends about them. There might be a person in your social circle who is catfishing you, and asking your friends can be super helpful. Search their picture on Google A Simple Google search can give you an insight about into a stranger is legit or not. If a person is talking to you online, you can save his or her and search it on Google. If the person is fake, you might see another name below the picture or find it online in many cases. The process is called reverse image search and it is easy for anyone to do it. Look for clinginess Most catfishers try to exploit people via dating apps like Tinder, Bumble or Mingle because they know people are vulnerable when it comes to love. This is where you should be wary of anything that smells fishy. If a stranger online is being too clingy and love struck, he or she is faking it in most cases. Further, if they confess love and emotions way too early, it is a big red flag. Video call them If you are talking to a stranger online for quite a few weeks now, ask them to a video call to see if he or she is real. In this way, you will find out the reality of your situation. However, if you are talking to a catfisher, he won’t agree for a video call or personal pictures and might share senseless stories. Don’t be afraid to ask them online their real name or where they work at. A real person won’t hide personal information! Believe your instincts Sometimes, you have to trust your basic instincts. As humans, we are hardwired to sense danger or trouble, you should go with your guts. If you think a stranger online is not what you imagine, the best route is to block them or ignore their texts. Why is Catfishing Still Common When People Know the Red Flags? Some people are lonely and depressed. They have no one to talk to, so catfishers try to befriend them online and exploit their loneliness and depression. The easiest way for catfishers and scammers to target is online dating apps. Besides people finding real love and relationships, many lonely people are just looking for someone to talk to and share their personal feelings. As we are getting restricted to our houses more, the usage of online dating is skyrocketing. Therefore, people are still being catfished because they are lonely and need company. Why Did People Start Catfishing Scams during the Pandemic? The COVID-19 pandemic forced people to stay in their homes and maintain social distancing. Many people were lonely and isolated (including me) and looking for something fun to do. Many people started creating NSFW groups and fake identities to fulfill their desires. Because there was nothing else to do at home, creating a fake identity and catfishing people was exciting for many people online. There was a story on Reddit where a catfisher confessed that he started catfishing people out of loneliness and boredom. The person remained anonymous and said that he created a fake NSFW account of a girl and talked to many desperate people online, including male, female, and other genders. It turns out loneliness and boredom is the root of all catfishing problems. Be Wary of Catfishing Scams We hope now you have a clear picture of what’s happening online and how people are being catfished. And, this guide can help you prevent any catfishing or online scams and make you aware of the online dangers.

How to Not Be a Victim to the Catfishing Pandemic Spike 

As communities worldwide grapple with the reality of an extended , Internet usage has, understandably, significantly increased. As more and more countries are forced into lock-down in an effort to curb the spread of the deadly virus, the growing number of people forced into working remotely and finding online entertainment is seeing the Internet absolutely explode. In fact COVID-19 has pushed up Internet use by a whopping 70 percent in some countries while streaming services are up by more than 12 percent, . COVID-19 induced lock-down figures last month revealed The Internet is getting overwhelmed. And, while Internet service providers (ISPs) like Comcast, Vodafone, Verizon and Telefónica have been building their networks for years to account for increased demand, they never anticipated it would happen quite so suddenly. Enrique Blanco, CTO at Telefónica, a Spanish telecommunications company, said that growth the industry had expected to take a year is “happening over days”, due to COVID-19. “Increased consumption of digital content from mobile apps to free TV streaming and gaming has already been observed in China and Italy,”a statement by Omdia, the tech research arm of Informa Tech, said. “Online streaming services provided by brands such as Netflix and now Disney+ are likely to see 12% growth, while Ecommerce will be the other sector that will see a revenue boost as a result of the pandemic, adding $175 billion in revenue in 2020, which represents a 5% increase.” But while growth is usually seen as a good thing, things in the digital space aren’t exactly golden. The Internet is buckling under the pressure of unprecedented heavy use, and some ISPs are . ISPs look after a number of very important Internet-related things including Internet access, Internet transit, domain name registration, web hosting, Usenet service, and colocation. In other words, ISPs are the guys tasked with making sure you can access the Internet whenever you like, as well as routing Internet traffic, resolving domain names, and maintaining the network infrastructure that makes Internet use possible. Without ISPs, there would be no Internet. But amidst staggering rates of Internet use, we are beginning to see service providers strain under the pressure. struggling to deal with the additional demand for connectivity In Italy, , Internet traffic increased by 109.3 percent over the roughly one-month period from February 20 to March 31, while download speed decreased by 35.4 percent. In India, the mean fixed broadband speed has been steadily declining since the beginning of lock-down too. one of the nation's worst hit by the global coronavirus pandemic "When networks are under usage strain like they are in this unprecedented time of lockdown in India due to COVID-19, it is natural that they experience some level of slowdown. It is important to note that while the Internet itself should handle elevated usage, there may be impacts to speed as people continue to move their daily activities increasingly online. While the core of the Internet remains stable, some ISP networks may struggle to keep up,” CEO of Ookla, a web service that provides free analysis of Internet access performance metrics, , said. As per the latest Ookla's Speedtest Global Index report, the mean download speeds on fixed broadband have decreased from 39.65 Mbps in February to 35.98 Mbps in March - a direct result of increased global Internet use. Doug Suttles A few weeks ago, five of Spain's major operators (including Orange and Vodafone) warned that services could potentially soon collapse amidst a 40 percent spike in traffic through IP networks and 50 percent hike in voice calls. At around the same time, Netflix recognised it couldn’t continue to keep up with the growing demands of viewers while not impacting other Internet capacities, and so in a bid to support the sudden growth in streaming. Apple TV+, Disney+ and Amazon have followed suit. it reduced its streaming bit rates Interestingly, ISPs are coping at varying levels depending on the location of traffic congestion. which looked at the Internet congestion of Australia's major cities from February up until March 30, exposed huge differences in the speed from city to-city. It found that Internet speed was strongest in Perth, while Canberra and Melbourne were experiencing the worst average Internet congestion, according to data supplied by Monash University. A recent study led by KASPR datahaus, The good news is there are . If Internet users are struggling to get speedy and efficient Internet, and they are approaching the end of a contract, it is possible to swap to a new provider free of charge. However doing that won’t necessarily guarantee enhanced quality of Internet speed. options in the ISP space The other good news is, current effects on Internet speed are entirely regional, and we have already seen the worst of it. Speeds are unlikely to decrease much more, as most places around the world have already been issued stay-at-home orders. In other words, the most extreme cases of speed degradation have already occurred.

As Internet Usage Spikes During COVID-19 Pandemic: How Are ISPs Holding Up?

Coronavirus outbreak has trembled the world with global . This outbreak has affected the world catastrophically. Scientists have galvanized in order to find a vaccine against this virus, and a lot of funding is being raised in this regard. This has proven to be a life-threatening virus, and with more deaths with every passing day, it's safe to say chaos is being created all across the globe. cases reaching 182,742 with death toll crossing 7,174 The Coronavirus outbreak is expected to cut the global economy by half. Flights to and from many countries are banned to avoid this viral spread. Various technologies are being deployed for employee attendance, passenger’s verification or boarding at the airports, contactless ID verification in the form of facial recognition to ensure security from fraud as well as from coronavirus simultaneously. Various Touchless Technologies Following are some of the technologies being deployed to provide security while restricting the spread of this deadly virus: Contactless Biometric Biometric fingerprint scanning used in immigration checkpoints at airports, offices, buildings, hospitals are the carriers of coronavirus and many people come in direct contact with them. Coronavirus is spread across 150 countries across the globe affecting more than 183,000 patients. The flights and business operations are grounded in fear of the spread of this virus. Fingerprint scanners carry different germs on them including viruses and bacteria. These scanners are considered contaminated and any direct physical contact with such scanners can transfer the virus to healthy ones becoming a reason for the widespread virus. Therefore are required for identification for security purposes as well as to stop the spread. ID verification at airports using contactless biometrics instead of fingerprint scanning can be very helpful. Facial recognition technology is a contactless form of biometric and verifies a person just like taking a selfie without coming in physical contact with any scanner or device. Contactless and palm vein features by generating image acquisition can also be used for identification purposes. Less-constrained biometric systems are of high interest for implementation at public places to identify people from a greater distance. contactless biometric systems Palm recognition Identity Verification of Masked People Advanced facial recognition technology is being used which helps to identify and verify the faces of people donning face masks at public squares. With high efficiency and precision rate, this technology is able to identify the faces of people wearing masks. This technology is used for many purposes including traveler’s ID verification during onboarding, attendance of staff, and even recognizing people not wearing masks as a preventive measure. y can help ensure security from fraud as well as from coronavirus simultaneously as no human interaction is required by deploying this technology. Facial recognition technolog Touchless Bag Drop for Luggage To simplify and streamline the baggage check-in process . In this, passengers use their reference number to print a boarding card or to weigh their luggage and print a tag. Once the luggage is tagged, it is taken to the self-service kiosk to automatically dispatch the luggage. Such a system is already in use at several airports around the world. This contactless technology has won praise from customers and staff for enhancing passenger’s experience and restricting the spread of the virus. touch-less bag drop is deployed at London Luton Airport Digital Payment Solutions Whether you are a business providing digital services to customers, running e-stores, any online business having the facility of online transactions is a must as cash is now the declared source of spreading COVID-19. The major responsibility is on digital banks, financial institutions, and e-commerce sites. So digital transactions are a must in this pandemic. First Canadian bank has recently introduced digital identity verification into their systems. Drones and Robot Nurses: Drones are being used to supply safety equipment and medicines to suspected people. By this technology no direct contact takes place between humans and the spread of this virus can be stopped. As this deadly disease is highly contagious, so to minimize human interactions robot nurses are being used in hospitals to provide necessary things to patients and also to disinfect the areas. COVID-19 has created several challenges for the world as it is reaching every corner of the world. Preventive measures need to be taken to stop its spread. in many ways. But the challenge of incorporating the coronavirus spread that happens because of contamination is that scanners at airports can be handled using contactless technologies to identify and ease people without spreading coronavirus. This is the time to adopt innovative touchless technologies for airports. Technology is helping to fight back coronavirus

Contactless Technology Spikes in Demand Amid Coronavirus

To get more information's about what is the Outbreak tool, read . this post Here i will discuss just how we can improve the algorithm, by analyzing the detection of an article as viral multiple times, for the same in multiple Fb Pages or within the same page. permanent_link As a study-case i will select this image post. The algorithm first detected this post when it had 1400 shares, and was growing at as a speed of 760 shares per hour. This is not a fake article. But is not true also. Welcome to Logical Fallacies. This specific post is a logical fallacy, called Argument from ignorance. I will let Wikipedia do the explanation: (from : ), also known as (in which represents “a lack of contrary evidence”), is a . It asserts that a proposition is true because it has not yet been proven false (or vice versa). Argument from ignorance Latin argumentum ad ignorantiam appeal to ignorance ignorance fallacy in informal logic This represents a type of in that it excludes a third option, which is that: there may have been an insufficient investigation, and therefore there is insufficient information to prove the proposition be either true or false. Nor does it allow the admission that the choices may in fact not be two (true or false), but may be as many as four, false dichotomy true false unknown between true or false being unknowable (among the first three). [1] In debates, appeals to ignorance are sometimes used in an attempt to shift the . burden of proof Also, if we look at the emotional fingerprint of the post, we can see a big difference compared with a fake news article. OK, now that we had finished explaining why this is a shady post, let`s look at the data. The algorithm detected the post 4 times in the last 24 hours, the first time when it had 1432 shares, then next hour when it was at 2142 shares, 2 hours after when it rose to 2875 shares and final time when it was at 3592 shares. This allows us to analyze recurring viral identified articles, and elevate them to a higher viral threat configuration. I`m doing a 5 levels classification of virality, but for the sake of simplicity, let`s explain it using the count method. We can add a column that does the count, if this article is declared viral, increase the count to 1. If we detect the article again, in the future hours, we increase the count to 2. Journalists can set a threshold and get in their feed only the articles that were detected 2 times as viral, so that they have less articles to look into. I don`t have a front End to the app, so the only way to output the results is via . The first results are the oldest, scroll to the end for the most recent future possible future fake news. this google spreadsheet In the list i have a list of grey listed and legitimate websites. I have added some white listed Fb pages, and you will see posts from that pages because i want to learn the difference between viral news that are legitimate compared with the ones that are fake or misleading. Also, i`m scanning some politicians fb pages, because they post news that can be related, and they can be the first to break a news. Help support this effort: https://www.youcaring.com/newspapersandjournalists-711969 Sorry for the language mistakes in the article, English is not my first language. About Me In the last 3 years i`m a collaborator with the , were i do data analysis and pattern recognition to uncover patterns of corruption in unstructured datasets. Organised Crime and Corruption Reporting Projects ( OCCRP ) In September 2016 i have moved to San Francisco, to start a new life here. You can find me online on Hackernoon , Medium Florin Badita, AngelList , Twitter , L inkedin, O penstreetmap , G ithub, Q uora, F acebook

The Outbreak: How to Detect The Real Viral Posts Compared to The One Hour Share Spike

Backed up by the data, made an announcement that marks a sudden surge in its value from to . CoinDesk Bitcoin $5000 $8101.91 Which is in fact a revolutionary spike considering the previous values of Bitcoin. The paradox of still remains to be a which token buyers and other skeptics are expecting to burst real soon. cryptocurrency bubble As per the record and , the concrete value of Bitcoin shot through the roof between to of Universal Time Zone ( ). Fast rewinding the recent history of Bitcoin trend, it looks like, after the trading, the Bitcoin value hit its rock bottom with $5000 as the closing value. conventional market analytics 20:00 20:15 UTC November 12 The submitted reports mention that the requisite reason for Bitcoin value hitting this low was when decided to call off its plan to upgrade the Bitcoin network. Segwit2x Early this November, some of the Bitcoin miners and Bitcoin communities a few, including Segwit2X development team announced that they’re backing off from supporting the . Bitcoin fork In order to raise the trust state amid the Bitcoin users and to keep the Bitcoin bubble stable, CoinDesk came up with a about its . Even other exchanges took similar actions to protect their consumers. Establishing themselves as some of the biggest gainers in this whole exercise. transparent description transaction process Having come this long, even now, there’s a raving opportunity for Bitcoin Exchanges, because of the and . rabid users investor interests A Bitcoin exchange is the best you can be right now, as the middlemen, who make the most out of every Bitcoin transaction, having a stable flow of income no matter what happens to the . BTC This is a once in a decade startup opportunity, to start a with , eating up with every madly driven spike. bitcoin exchange wallet service 2.5% transaction fees Probably amounting to millions a week!

Whopping 47% Spike In Bitcoin Value!

Remember my recent article ? “What an in-memory database is and how it persists data efficiently” In that article, I summarized the mechanisms that an in-memory adopts to persist data. I was talking about two mechanisms: transaction logging and snapshotting. I covered transaction logging (only in general, though) and touched snapshotting a bit. So, in this article I’m going to elaborate on snapshotting. I’ll start with a simple, quick & dirty implementation of snapshotting in an in-memory database, point out some problems of that and then dive deeply into the implementation of snapshotting in . database Tarantool Ok. We have an in-memory database. Everything is in memory. Right? As I mentioned in my previous , to snapshot it, we just need to write everything to a file. This means that we need to iterate through all the tables and through all the rows of all the tables and dump everything to disk in a single file via the ‘write’ syscall. Sounds simple. But the problem here is that the database is being changed. Even if we lock data structures as we go, we can end up with an inconsistent state on disk. article How to make it consistent? The easiest (and the dirtiest) way is to lock the whole database in advance, dump it to a file and unlock it. Well, this will work. A database can be locked for a long time. For example, if the size of the dataset is 256Gb, then with an HDD disk and its peak performance of 100Mb per second, the snapshotting will last for 256Gb/100Mb that equals (roughly) 2560 seconds or (roughly) 40 minutes. The database will still respond to queries but it will be all locked for update operations for 40 minutes. Dude, are you serious? (you’re saying right now). Well, 40 minutes of update downtime for a database, say, per day means 97% uptime in the very best case scenario (which, frankly, will never happen because there are numerous of other conditions to make your database go down). Any ideas? Let’s look closer at what is happening here. We locked the whole dataset only because we had to copy it to a slow device. What if we sacrifice memory to improve the speed? I mean, what if we copy the whole dataset in a separate place in memory and then dump it to a slow disk? Well, that sounds better, but entails at least three problems (big & small): 1. We still need to lock the whole dataset. Let’s say that we’re able to copy the dataset in memory at the rate of 1Gb per second (which still sounds optimistic because in practice the rate could be 200–500Mb per second for more of less sophisticated in-memory data structures). 256Gb/1Gb == 256 == (roughly) 4 minutes. 4 minutes of update downtime per day or 99.7% uptime, which is definitely better than 97%, but still not good enough. 2. After we have the whole dataset copied to a separate buffer in RAM, we need to dump it to disk. While we’re dumping it, the original copy of a database is changing. We need to handle it somehow, e.g. to save the transaction id with the snapshot to keep track of which transaction was the last considered within the snapshot. It’s actually no big deal, but still needs to be done. 3. We double down on RAM. I mean that we really need twice as much memory as we need for just a dataset, ALWAYS. Not only during snapshotting, but ALWAYS, because you can’t just add memory to a machine only during snapshotting and then take it back. One of the ways to fix this problem is to use use system provided by . When you fork, then a separate process is created with its own virtual address space, which has a read-only copy of the whole dataset. Read-only, because all changes happen in a parent process. So, we fork a process and then take our time to dump the dataset to disk. The only question here is — how is it different from the previous algorithm? The answer lies in the mechanism of Linux. COW is an acronym for “copy on write”. What does that mean? It means that a child process initially shares all the with its parent. And then, as soon as a parent or a child changes something in memory, a copy of a page is created. copy-on-write mechanism fork COW memory pages Of course, this copying entails some delays in responding, because we need to copy the page and even more. The size of a page is normally 4Kb. If you, say, change a small value in a database, then a will be raised first (because all the pages of both the parent and the child are marked as read-only after fork), then the system will switch to the kernel mode, then allocate a new page, copy 4Kb from the old page to the new one, and return back to the user mode. This is still a very simplified description of what is happening. For more details, start from here: . If this change touches many pages (which is very likely for spacious data structures like trees), then the same thing will happen many times. So, you can really slow down a database due to the COW mechanism. For heavy workloads, this delay can result in bad latency spikes or even in a short downtime. Plus, for heavy workloads, there will be numerous random page updates in the parent process, which can make almost the whole database be copied, which in turn can entail a double down on RAM. So, if you’re lucky, then no latency spikes, no downtime. If not, then yes to latency spikes and yes to downtime, oh and yes to double down on RAM. page fault exception https://en.wikipedia.org/wiki/Kernel_same-page_merging Another problem with fork is that it copies the . If you use, say, 256Gb of memory then the size of this table could be hundreds of megabytes, and your process can freeze for a second or so, which entails latency spikes. table of page descriptors Fork is not the ideal solution. Right? But this is the best that we’ve come up with so far. Actually, some popular in-memory databases are still using fork for snapshotting. For example, . Redis How to make a better solution? Let’s look closer at the COW mechanism. It “cows” by 4Kb. So, you change only one byte, but still “cow” the whole page (in fact, many pages when it comes to trees, even if you don’t need rebalancing). What if we just implement our own COW that will copy-on-write only the pieces of memory that were actually changed? More specifically, only the that were changed. Of course, we won’t do that as a substitution for the system mechanism, but just for our snapshotting needs. values The main idea here is to augment all our data structures (tree, hashes, table spaces) with the ability to store many versions of each data item. That is very close to . The difference is that it is used here not for concurrency control in general, but just for snapshotting. When snapshotting starts, all change operations now and on create newer versions of data items, whilst all older versions are still alive and used in snapshotting. Look at these pictures. This logic applies to trees, hashes and table spaces: multi-version concurrency control (MVCC) You see that data items can have old versions and new versions. For example, at the last picture you see that values 3,4,5 and 8 of the table space have old versions along with new versions attached to them, while values 1,2,6,7,9 have only single versions. Only newer versions are being changed. The older ones are used for read operations during snapshotting. The main difference between this MVCC-like COW and the system COW is that we “cow” not the whole 4Kb page, but only a small piece of data that is actually changing. Say, if you update a 4-byte integer number, then this will create a new version of this 4-byte integer and only these 4 bytes will be copied (plus some bytes as overhead for versioning). Compare this with 4096 bytes that were copied during the system copy-on-write, plus page faults and context switching (every context switching actually costs like copying 1Kb of memory or so), plus repeat all of these several times. Lots of job? Only to change a 4-byte integer during snapshotting. We adopted this manual COW technique for snapshotting in since 1.6.6 version (before that we used fork). Tarantool There will be more articles about that with more details and graphs from Mail.Ru Group’s production servers. Stay tuned.

How to avoid latency spikes and memory consumption spikes during snapshotting in an in-memory…

In my latest letter, Sell in May and Go Away, I highlighted that we were heading for the low liquidity season of the year and that the crypto markets would need a pause before trending higher. At this moment in time, I think the sluggish summer may finally be coming to an end. Uncertainty regarding the U.S. elections is in front of us, and given the many events that unfolded this month, I think we can expect plenty more volatility where investors will be “trading the election” until we see the results in November. Trump Going to the Oval Office? Trump is likely to become the next president. Following the assassination attempt, prediction markets put his likelihood at 71%, and now with Biden’s withdrawal, that has fallen to a comfortable 60%, with Kamala Harris looking like the second most obvious choice with a 38% probability currently. Digital Assets is the Talk of Town What’s interesting about this election is that digital assets have become a topic for candidates to discuss. Apparently, the political elite has finally understood that the +50M crypto holders in America represent a big voter base. This weekend, Trump and R. Kennedy Jr. are both speaking at the Bitcoin2024conference. Kamala Harris was also expected to speak there but declined the opportunity just a few days ago.Another interesting aspect is Trump’s choice of running mate, JD Vance. JD Vance is a self-proclaimed BTC holder, has been a strong proponent of the crypto industry in the Senate, and one of Gary Gensler's most vocal critics in recent years. A bitcoiner on the running card. How about that? Is it Finally Time to Lower the Rates? It’s now been a year since the Fed raised its benchmark rate to between 5.25% and 5.5%, holding this target range steady to date. In the meantime, Jerome Powell has said again and again that rates would not be lowered before something would either break or before the Fed would have been convinced inflation rates would trend lower towards the target rate of 2%. The most recent economic data signals that the Fed could soon start cutting rates: The Fed is getting closer to its 2% inflation target. In June, the monthly CPI fell by 0.1%, while the annual CPI rose by 3%. The U.S. economy is finally softening. GDP growth slowed to an annual rate of 1.4% in Q1 2024, down from 3.4% in the previous quarter. The labor market is also cooling down; June’s unemployment rate spiked to 4.1%, the highest it has been since November 2021. Additionally, Jerome Powell appeared on Capitol Hill this month and warned that “reducing policy restraint too late or too little could unduly weaken economic activity and employment.” Is the “higher for longer” regime coming to an end? It is still quite a balancing act for the Fed, but it seems inevitable that something has to move sooner rather than later. Currently, the market is predicting a 95.3% chance of a rate cut in September’s FOMC meeting. Bitcoin Remains Resilient Despite Sell-offs Bitcoin’s price has remained range-bound between $60K and $70K for most of the summer despite the drying liquidity from Bitcoin ETFs, which had propelled the price from lows of $42K before the approval to record highs of $73K in March. But what’s more remarkable is BTC’s resilience following the recent $2.87 billion sell-off by Germany’s Saxony state. That said, there's a risk of additional selling pressure. Mt. Gox, the defunct crypto exchange, has started its repayment plan, which aims to reimburse creditors around $9 billion worth of Bitcoin and Bitcoin Cash. 140,000 BTC; this is worth over $9 billion at the current market price. So far, over 40% of the repayments have already been distributed to the creditors. On the brighter side, repayments in-kind does not equal selling pressure as creditors may decide to HODL their BTC. On-chain data shows that BTC withdrawals on Kraken spiked after the exchange completed distributing its share of allocated repayments. This is a sign that a good number of Mt. Gox creditors could be moving their repayments to cold wallets for long-term HODLing. Personally, I remain optimistic that BTC’s price will trend higher in H2 based on its current resilience amid the much-awaited Mt. Gox repayments, which most had predicted would be a huge sell-off event. Additionally, there are some macro and industry-specific factors that could be catalysts in sustaining the current price levels or pushing BTC higher: A good number of the creditors might decide to HODL onto their BTC, causing only short-term downward pressure on BTC’s price. Once the uncertainty over Mt. Gox repayments is over, there is no big expected selling pressure. There is a high chance that the Fed will cut rates in September, likely triggering a rally in risk-on assets. Liquidity is gradually returning to the market. Blackrock’s Bitcoin Spot ETF recently recorded a daily net inflow of $526 million, the largest since March 13th. The crypto market could soon play catch-up with the S&P 500, Nasdaq, and Gold, which showed strength in Q2 while BTC lagged over the past three months. The crypto markets are favoring the Republican Party, and currently, it remains likely that Trump will take office. In my latest letter, Sell in May and Go Away , I highlighted that we were heading for the low liquidity season of the year and that the crypto markets would need a pause before trending higher. At this moment in time, I think the sluggish summer may finally be coming to an end. Sell in May and Go Away Sell in May and Go Away Sell in May and Go Away Uncertainty regarding the U.S. elections is in front of us, and given the many events that unfolded this month, I think we can expect plenty more volatility where investors will be “trading the election” until we see the results in November. Trump Going to the Oval Office? Trump is likely to become the next president. Following the assassination attempt, prediction markets put his likelihood at 71%, and now with Biden’s withdrawal, that has fallen to a comfortable 60%, with Kamala Harris looking like the second most obvious choice with a 38% probability currently. Digital Assets is the Talk of Town What’s interesting about this election is that digital assets have become a topic for candidates to discuss. Apparently, the political elite has finally understood that the +50M crypto holders in America represent a big voter base. +50M crypto holders +50M crypto holders This weekend, Trump and R. Kennedy Jr. are both speaking at the Bitcoin2024 conference. Kamala Harris was also expected to speak there but declined the opportunity just a few days ago. Another interesting aspect is Trump’s choice of running mate, JD Vance. JD Vance is a self-proclaimed BTC holder, has been a strong proponent of the crypto industry in the Senate, and one of Gary Gensler's most vocal critics in recent years. Bitcoin2024 A bitcoiner on the running card. How about that? Is it Finally Time to Lower the Rates? It’s now been a year since the Fed raised its benchmark rate to between 5.25% and 5.5%, holding this target range steady to date. In the meantime, Jerome Powell has said again and again that rates would not be lowered before something would either break or before the Fed would have been convinced inflation rates would trend lower towards the target rate of 2%. The most recent economic data signals that the Fed could soon start cutting rates: could The Fed is getting closer to its 2% inflation target. In June, the monthly CPI fell by 0.1%, while the annual CPI rose by 3%. The Fed is getting closer to its 2% inflation target. In June, the monthly CPI fell by 0.1%, while the annual CPI rose by 3%. The U.S. economy is finally softening. GDP growth slowed to an annual rate of 1.4% in Q1 2024, down from 3.4% in the previous quarter. The U.S. economy is finally softening. GDP growth slowed to an annual rate of 1.4% in Q1 2024, down from 3.4% in the previous quarter. The labor market is also cooling down; June’s unemployment rate spiked to 4.1%, the highest it has been since November 2021. The labor market is also cooling down; June’s unemployment rate spiked to 4.1%, the highest it has been since November 2021. The labor market is also cooling down; June’s unemployment rate spiked to 4.1%, the highest it has been since November 2021. Additionally, Jerome Powell appeared on Capitol Hill this month and warned that “ reducing policy restraint too late or too little could unduly weaken economic activity and employment. ” reducing policy restraint too late or too little could unduly weaken economic activity and employment. Is the “higher for longer” regime coming to an end? It is still quite a balancing act for the Fed, but it seems inevitable that something has to move sooner rather than later. Currently, the market is predicting a 95.3% chance of a rate cut in September’s FOMC meeting. Bitcoin Remains Resilient Despite Sell-offs Bitcoin’s price has remained range-bound between $60K and $70K for most of the summer despite the drying liquidity from Bitcoin ETFs, which had propelled the price from lows of $42K before the approval to record highs of $73K in March. But what’s more remarkable is BTC’s resilience following the recent $2.87 billion sell-off by Germany’s Saxony state . Germany’s Saxony state Germany’s Saxony state That said, there's a risk of additional selling pressure. Mt. Gox, the defunct crypto exchange, has started its repayment plan, which aims to reimburse creditors around $9 billion worth of Bitcoin and Bitcoin Cash. 140,000 BTC; this is worth over $9 billion at the current market price. So far, over 40% of the repayments have already been distributed to the creditors. On the brighter side, repayments in-kind does not equal selling pressure as creditors may decide to HODL their BTC. On-chain data shows that BTC withdrawals on Kraken spiked after the exchange completed distributing its share of allocated repayments. This is a sign that a good number of Mt. Gox creditors could be moving their repayments to cold wallets for long-term HODLing. Personally, I remain optimistic that BTC’s price will trend higher in H2 based on its current resilience amid the much-awaited Mt. Gox repayments, which most had predicted would be a huge sell-off event. Additionally, there are some macro and industry-specific factors that could be catalysts in sustaining the current price levels or pushing BTC higher: A good number of the creditors might decide to HODL onto their BTC, causing only short-term downward pressure on BTC’s price. A good number of the creditors might decide to HODL onto their BTC, causing only short-term downward pressure on BTC’s price. Once the uncertainty over Mt. Gox repayments is over, there is no big expected selling pressure. Once the uncertainty over Mt. Gox repayments is over, there is no big expected selling pressure. There is a high chance that the Fed will cut rates in September, likely triggering a rally in risk-on assets. There is a high chance that the Fed will cut rates in September, likely triggering a rally in risk-on assets. Liquidity is gradually returning to the market. Blackrock’s Bitcoin Spot ETF recently recorded a daily net inflow of $526 million, the largest since March 13th. Liquidity is gradually returning to the market. Blackrock’s Bitcoin Spot ETF recently recorded a daily net inflow of $526 million, the largest since March 13th. The crypto market could soon play catch-up with the S&P 500, Nasdaq, and Gold, which showed strength in Q2 while BTC lagged over the past three months. The crypto market could soon play catch-up with the S&P 500, Nasdaq, and Gold, which showed strength in Q2 while BTC lagged over the past three months. The crypto markets are favoring the Republican Party, and currently, it remains likely that Trump will take office. The crypto markets are favoring the Republican Party, and currently, it remains likely that Trump will take office.

Here's Why It’s Time to Wake Up Again…

Lumoz: A New Paradigm for Rollup Platformization

লুমোজ: রোলআপ প্ল্যাটফর্মাইজেশনের জন্য একটি নতুন দৃষ্টান্ত

Lumoz: Ein neues Paradigma für die Rollup-Plattformisierung

Lumoz: un nuevo paradigma para la plataforma acumulativa

Lumoz : un nouveau paradigme pour la plateforme de rollup

लुमोज़: रोलअप प्लेटफ़ॉर्माइज़ेशन के लिए एक नया प्रतिमान

Lumoz: ロールアップ プラットフォーム化の新しいパラダイム

Lumoz: 롤업 플랫폼화를 위한 새로운 패러다임

Lumoz: um novo paradigma para plataforma rollup

Lumoz: новая парадигма накопительной платформы

Lumoz: Toplama Platformizasyonu için Yeni Bir Paradigma

Lumoz: Một mô hình mới cho nền tảng cuộn lên

Lumoz：Rollup 平台化的新范式

The first half of 2024 has seen major enterprise SaaS players like Salesforce reporting declining sales, reflecting the broader challenges businesses have collectively faced this year. Working against influences like high interest rates, the economic uncertainty of an election year, and existential questions surrounding the future impacts of AI, IT leaders are tightening their budgets. Amidst such challenges, strategic partnerships play crucial roles in opening new revenue streams and enhancing software offerings – even longtime rivals Apple and Meta have recently flirted with the idea. Microsoft has long been an exemplar of this strategy, going so far as to claim that 95% of its commercial revenue is generated through partnerships. A recent Forrester study supports this, with over half of surveyed companies reporting that partnerships generate 20% of their overall revenue. The most lucrative partnerships are those that have invested time, money, and resources into comprehensive enablement processes including training, tools, knowledge, and support. But resource limitations dictate that not all partners are treated equally when it comes to enablement. Consequently, larger, top-tier partners get personalized, often hands-on support, while smaller, long-tail partners are more likely to be directed toward self-service portals or waiting on unfulfilled IT support tickets. This disparity hinders growth for both partners. Luckily, generative AI makes it possible to provide tier-one treatment to long-tail partners affordably. Here’s how. Personalize and Precipitate Partner Onboarding These smaller partners are often directed to generic portals containing vast amounts of documentation. They are expected to independently navigate this information overload to learn the necessary skills for selling and implementing solutions. This can be a significant barrier to entry, THUS Long-tail partners take considerably longer to become productive due to the time required to learn from impersonal resources. On average, students forget 70 percent of what we teach within 24 hours of the training experience. Companies need to bear in mind customizability and also focus on developing training programs that will help organizations achieve specific goals tailored to their unique needs. According to a McKinsey report, employees spend 1.8 hours every day—9.3 hours per week, on average—searching and gathering information. Increase Implementation and Sales It’s estimated that around 80 percent of product features are “rarely or never used.” Without proper guidance, long-tail partners may struggle to grasp the nuances of the software, hindering their ability to convert leads and deliver successful implementations. On average, young companies (1-2 years old) start out with 29 SaaS apps. By the time they’re 3-6 years old, that number spikes to 103 apps. (Bettercloud)


With so many apps, most are NOT using them to max capacity


Over 50% of SaaS licenses are unused for more than 90 days. (Redline) because people don’t know how to use products!



Dashboards can illuminate which features are being underutilized that if properly deployed could lead to improved results Decrease Churn 54% of businesses want to optimize and find savings in their software spend. (Flexera)


Don’t be the area that gets cut due to inefficiencies



25% of IT teams say that they spend the majority of their time managing third-party vendors and business solutions. (Productiv) AI may not yet be ready to be an assistant taking on our working tasks, but it is perfectly suited as an agent that can cater to the needs of long-tail partners. The first half of 2024 has seen major enterprise SaaS players like Salesforce reporting declining sales, reflecting the broader challenges businesses have collectively faced this year. Working against influences like high interest rates, the economic uncertainty of an election year, and existential questions surrounding the future impacts of AI, IT leaders are tightening their budgets. The first half of 2024 has seen major enterprise SaaS players like Salesforce reporting declining sales , reflecting the broader challenges businesses have collectively faced this year. Working against influences like high interest rates, the economic uncertainty of an election year, and existential questions surrounding the future impacts of AI, IT leaders are tightening their budgets. declining sales declining sales Amidst such challenges, strategic partnerships play crucial roles in opening new revenue streams and enhancing software offerings – even longtime rivals Apple and Meta have recently flirted with the idea . Microsoft has long been an exemplar of this strategy, going so far as to claim that 95% of its commercial revenue is generated through partnerships. flirted with the idea flirted with the idea 95% of its commercial revenue 95% of its commercial revenue A recent Forrester study supports this, with over half of surveyed companies reporting that partnerships generate 20% of their overall revenue. Forrester study Forrester study The most lucrative partnerships are those that have invested time, money, and resources into comprehensive enablement processes including training, tools, knowledge, and support. But resource limitations dictate that not all partners are treated equally when it comes to enablement. Consequently, larger, top-tier partners get personalized, often hands-on support, while smaller, long-tail partners are more likely to be directed toward self-service portals or waiting on unfulfilled IT support tickets. This disparity hinders growth for both partners. Luckily, generative AI makes it possible to provide tier-one treatment to long-tail partners affordably. Here’s how. Personalize and Precipitate Partner Onboarding Personalize and Precipitate Partner Onboarding These smaller partners are often directed to generic portals containing vast amounts of documentation. They are expected to independently navigate this information overload to learn the necessary skills for selling and implementing solutions. This can be a significant barrier to entry, THUS Long-tail partners take considerably longer to become productive due to the time required to learn from impersonal resources. These smaller partners are often directed to generic portals containing vast amounts of documentation. They are expected to independently navigate this information overload to learn the necessary skills for selling and implementing solutions. This can be a significant barrier to entry, THUS Long-tail partners take considerably longer to become productive due to the time required to learn from impersonal resources. On average, students forget 70 percent of what we teach within 24 hours of the training experience. On average, students forget 70 percent of what we teach within 24 hours of the training experience. forget 70 percent forget 70 percent Companies need to bear in mind customizability and also focus on developing training programs that will help organizations achieve specific goals tailored to their unique needs. Companies need to bear in mind customizability and also focus on developing training programs that will help organizations achieve specific goals tailored to their unique needs. According to a McKinsey report, employees spend 1.8 hours every day—9.3 hours per week, on average—searching and gathering information. According to a McKinsey report, employees spend 1.8 hours every day —9.3 hours per week, on average—searching and gathering information. 1.8 hours every day 1.8 hours every day Increase Implementation and Sales Increase Implementation and Sales It’s estimated that around 80 percent of product features are “rarely or never used.” It’s estimated that around 80 percent of product features are “rarely or never used.” 80 percent 80 percent Without proper guidance, long-tail partners may struggle to grasp the nuances of the software, hindering their ability to convert leads and deliver successful implementations. Without proper guidance, long-tail partners may struggle to grasp the nuances of the software, hindering their ability to convert leads and deliver successful implementations. On average, young companies (1-2 years old) start out with 29 SaaS apps. By the time they’re 3-6 years old, that number spikes to 103 apps. (Bettercloud)


With so many apps, most are NOT using them to max capacity


Over 50% of SaaS licenses are unused for more than 90 days. (Redline) because people don’t know how to use products! Dashboards can illuminate which features are being underutilized that if properly deployed could lead to improved results On average, young companies (1-2 years old) start out with 29 SaaS apps. By the time they’re 3-6 years old, that number spikes to 103 apps. ( Bettercloud ) With so many apps, most are NOT using them to max capacity


Over 50% of SaaS licenses are unused for more than 90 days. (Redline) because people don’t know how to use products! Bettercloud Bettercloud With so many apps, most are NOT using them to max capacity Over 50% of SaaS licenses are unused for more than 90 days. (Redline) because people don’t know how to use products! With so many apps, most are NOT using them to max capacity With so many apps, most are NOT using them to max capacity Over 50% of SaaS licenses are unused for more than 90 days. (Redline) because people don’t know how to use products! Over 50% of SaaS licenses are unused for more than 90 days. ( Redline ) because people don’t know how to use products! Redline Redline Dashboards can illuminate which features are being underutilized that if properly deployed could lead to improved results Decrease Churn Decrease Churn 54% of businesses want to optimize and find savings in their software spend. (Flexera)


Don’t be the area that gets cut due to inefficiencies 25% of IT teams say that they spend the majority of their time managing third-party vendors and business solutions. (Productiv) 54% of businesses want to optimize and find savings in their software spend. ( Flexera ) Don’t be the area that gets cut due to inefficiencies Flexera Flexera Don’t be the area that gets cut due to inefficiencies Don’t be the area that gets cut due to inefficiencies Don’t be the area that gets cut due to inefficiencies 25% of IT teams say that they spend the majority of their time managing third-party vendors and business solutions. ( Productiv ) Productiv Productiv AI may not yet be ready to be an assistant taking on our working tasks, but it is perfectly suited as an agent that can cater to the needs of long-tail partners.

Playing the Long Game: How AI Can Optimize Enablement in Longtail Software Partnerships

Digital twins are highly accurate virtual versions of real-life places or things. They have become widespread in industries such as manufacturing because decision-makers want to test factory layouts, product prototypes or other specifics before approving them. Some leaders also build digital twins of critical equipment to detect when a machine’s operating conditions deviate too much from the virtual version’s norms, indicating an urgent problem to investigate. How can cybersecurity practitioners apply digital twins to their defensive strategies? Choose the Digital Twin’s Scope Digital twins are more involved than some people realize. Those interested in using them will get the best results by understanding what proposed digital twins will help them monitor. Digital twins have a three-layer architecture with edge servers, analytics software, and more. More specifically, there are hardware, software, and middleware elements. The hardware layer includes elements such as IoT sensors and routers, while the middleware components span connectivity, data processing and other essential functions. Finally, the software layer could feature machine learning modules, data dashboards or simulation tools, among others. Digital twins are highly customizable and will have various functionalities according to an organization’s or client’s needs. An early step in creating the digital twin must involve determining its scope. Will the digital twin represent an entire corporate network or specific, highly critical internet-connected devices? Answering that question allows people to assess related necessities about how the digital twin will work and its role in strengthening cybersecurity. Apply the Digital Twin to Threat Intelligence New and emerging cyberthreats keep internet security practitioners proactive, consistently trying to stay ahead of the bad actors. That can be a challenging aim, but digital twins can make it easier. One example comes from a company that has spent a decade capturing threat details. People from that enterprise gather information from technical sources, the internet, and the dark web, using it to help cybersecurity professionals recognize and prepare for the attacks that could compromise their networks. Part of the digital twin includes information about topics that have recently caught people’s attention and dominated their online conversations. Executives hope the insights will improve cybersecurity practitioners’ workflows, allowing them to work faster and more effectively. Additionally, users can become aware of potential threats faster, allowing them to respond to and mitigate them more quickly than they could without the digital twin. Digital twins are versatile, allowing people to update them as conditions change. Analysts believe their market worth will reach $154 billion by 2030. The ability to apply them to various situations is one reason for the anticipated continuation of these tools’ already widespread popularity. As the cybersecurity landscape evolves, digital twins can respond similarly by reflecting the latest threats in their stored data. Use the Digital Twin for Anomaly Detection Spotting potential network intrusion attempts is not easy, but digital twins can supplement human efforts. One possibility is to build a digital twin to reflect normal, expected network activity. Then, set it up to flag anything that strays from such conditions. Financial institutions take a similar approach by gathering data on individual customers and automatically blocking transactions that are too far outside of how, when, why, and where those people spend. Such applications explain why someone may need to contact a credit card provider before purchasing a big-ticket item that is out of character for them. There are three types of anomalies to detect as cybersecurity practitioners. The first is a point anomaly, which is a single unexpected event. Then, there are collective anomalies, which only stand out as abnormal when viewed in a group, when people may spot unusual patterns. Finally, contextual anomalies are unexpected events that appear out of context relative to the baseline. Suppose a cybersecurity team learns a site’s traffic is lowest in the early morning. If activity suddenly spiked during that time, they would know to investigate the cause. People can build digital twins to support their anomaly detection efforts, increasing their chances of knowing about unusual events sooner. Quick action reduces the chances of infiltrators entering networks unnoticed and wreaking havoc for days or weeks before someone discovers them. Select Appropriate Goals What do people want to achieve by building a digital twin for cybersecurity? They must determine that to keep projects on track for completion and make the outcomes as impactful as possible. One example comes from a project where people will build a digital twin of an electric grid. Those involved will use it to mitigate cascading failures and create proactive defense mechanisms. Additionally, they believe the digital twin will improve cybersecurity by boosting resilience and facilitating faster recoveries from outages. Decision-makers often use digital twins to run various scenarios in controlled, virtual environments. Seeing the ripple effects of those events allows users to consider how they would respond if the scenarios occurred in real life. The electric grid’s digital twin will predict potential threats, improving preparedness. Additionally, this four-year project will allow participants to test new applications that could expand digital twin utilization for power companies and other industries. Cybersecurity experts warn that critical infrastructure operators are prime targets for cybercriminals since such attackers want to cause the most significant disruptions for the largest number of people. Identify the Biggest Risks With the Digital Twin Cybersecurity professionals frequently use frameworks that detail best practices and reveal whether the respective organizations are doing the right things or need to improve. However, learning about the most prominent risks affecting particular industries and building digital twins to reduce them are also good strategies. Such was the case when an agency within the United States Department of Defense contracted a service provider to build a proof-of-concept digital twin to reduce cyberattacks on a small manufacturing system. The tool will rely on synthetic data about potential cyberattack vectors, enabling users to identify and defend against infiltrations before they affect the organization’s operational technology. The project’s participants believe their work will improve awareness and protection measures, making it harder for attackers to target the organization successfully. Since cybersecurity risks can vary depending on factors such as an entity’s industry, resources and size, people should strongly consider creating digital twins that can help them prepare for known threats and unfamiliar challenges that could arise soon. People can consider adopting this approach by using digital twins to learn which issues would cause the most damage if not addressed in time. Then, they can adjust their cybersecurity measures to increase organizational readiness for those potentially disruptive problems, reducing the likelihood of them causing problems in real life. Digital Twins Are Worth Exploring These examples and tips should inspire people interested in applying digital twins for tighter cybersecurity. In addition to following the suggestions here, individuals should consider how other industry peers use these tools for better cyberattack protection. Digital twins are highly accurate virtual versions of real-life places or things. They have become widespread in industries such as manufacturing because decision-makers want to test factory layouts, product prototypes or other specifics before approving them. Digital twins are highly accurate virtual versions of real-life places or things. They have become widespread in industries such as manufacturing because decision-makers want to test factory layouts, product prototypes or other specifics before approving them. Some leaders also build digital twins of critical equipment to detect when a machine’s operating conditions deviate too much from the virtual version’s norms, indicating an urgent problem to investigate. How can cybersecurity practitioners apply digital twins to their defensive strategies? Choose the Digital Twin’s Scope Digital twins are more involved than some people realize. Those interested in using them will get the best results by understanding what proposed digital twins will help them monitor. Digital twins have a three-layer architecture with edge servers, analytics software, and more. More specifically, there are hardware, software, and middleware elements. a three-layer architecture a three-layer architecture The hardware layer includes elements such as IoT sensors and routers, while the middleware components span connectivity, data processing and other essential functions. Finally, the software layer could feature machine learning modules, data dashboards or simulation tools, among others. Digital twins are highly customizable and will have various functionalities according to an organization’s or client’s needs. An early step in creating the digital twin must involve determining its scope. Will the digital twin represent an entire corporate network or specific, highly critical internet-connected devices? Answering that question allows people to assess related necessities about how the digital twin will work and its role in strengthening cybersecurity. Apply the Digital Twin to Threat Intelligence New and emerging cyberthreats keep internet security practitioners proactive, consistently trying to stay ahead of the bad actors. That can be a challenging aim, but digital twins can make it easier. One example comes from a company that has spent a decade capturing threat details. People from that enterprise gather information from technical sources, the internet, and the dark web, using it to help cybersecurity professionals recognize and prepare for the attacks that could compromise their networks. has spent a decade has spent a decade Part of the digital twin includes information about topics that have recently caught people’s attention and dominated their online conversations. Executives hope the insights will improve cybersecurity practitioners’ workflows, allowing them to work faster and more effectively. Additionally, users can become aware of potential threats faster, allowing them to respond to and mitigate them more quickly than they could without the digital twin. Digital twins are versatile, allowing people to update them as conditions change. Analysts believe their market worth will reach $154 billion by 2030. The ability to apply them to various situations is one reason for the anticipated continuation of these tools’ already widespread popularity. As the cybersecurity landscape evolves, digital twins can respond similarly by reflecting the latest threats in their stored data. market worth will reach $154 billion market worth will reach $154 billion Use the Digital Twin for Anomaly Detection Spotting potential network intrusion attempts is not easy, but digital twins can supplement human efforts. One possibility is to build a digital twin to reflect normal, expected network activity. Then, set it up to flag anything that strays from such conditions. Financial institutions take a similar approach by gathering data on individual customers and automatically blocking transactions that are too far outside of how, when, why, and where those people spend. Such applications explain why someone may need to contact a credit card provider before purchasing a big-ticket item that is out of character for them. There are three types of anomalies to detect as cybersecurity practitioners. The first is a point anomaly, which is a single unexpected event. Then, there are collective anomalies, which only stand out as abnormal when viewed in a group, when people may spot unusual patterns. three types of anomalies to detect three types of anomalies to detect Finally, contextual anomalies are unexpected events that appear out of context relative to the baseline. Suppose a cybersecurity team learns a site’s traffic is lowest in the early morning. If activity suddenly spiked during that time, they would know to investigate the cause. People can build digital twins to support their anomaly detection efforts, increasing their chances of knowing about unusual events sooner. Quick action reduces the chances of infiltrators entering networks unnoticed and wreaking havoc for days or weeks before someone discovers them. Select Appropriate Goals What do people want to achieve by building a digital twin for cybersecurity? They must determine that to keep projects on track for completion and make the outcomes as impactful as possible. One example comes from a project where people will build a digital twin of an electric grid. Those involved will use it to mitigate cascading failures and create proactive defense mechanisms. Additionally, they believe the digital twin will improve cybersecurity by boosting resilience and facilitating faster recoveries from outages. mitigate cascading failures and create mitigate cascading failures and create Decision-makers often use digital twins to run various scenarios in controlled, virtual environments. Seeing the ripple effects of those events allows users to consider how they would respond if the scenarios occurred in real life. The electric grid’s digital twin will predict potential threats, improving preparedness. Additionally, this four-year project will allow participants to test new applications that could expand digital twin utilization for power companies and other industries. Cybersecurity experts warn that critical infrastructure operators are prime targets for cybercriminals since such attackers want to cause the most significant disruptions for the largest number of people. Identify the Biggest Risks With the Digital Twin Cybersecurity professionals frequently use frameworks that detail best practices and reveal whether the respective organizations are doing the right things or need to improve. However, learning about the most prominent risks affecting particular industries and building digital twins to reduce them are also good strategies. Such was the case when an agency within the United States Department of Defense contracted a service provider to build a proof-of-concept digital twin to reduce cyberattacks on a small manufacturing system. The tool will rely on synthetic data about potential cyberattack vectors, enabling users to identify and defend against infiltrations before they affect the organization’s operational technology. a proof-of-concept digital twin a proof-of-concept digital twin The project’s participants believe their work will improve awareness and protection measures, making it harder for attackers to target the organization successfully. Since cybersecurity risks can vary depending on factors such as an entity’s industry, resources and size, people should strongly consider creating digital twins that can help them prepare for known threats and unfamiliar challenges that could arise soon. People can consider adopting this approach by using digital twins to learn which issues would cause the most damage if not addressed in time. Then, they can adjust their cybersecurity measures to increase organizational readiness for those potentially disruptive problems, reducing the likelihood of them causing problems in real life. Digital Twins Are Worth Exploring These examples and tips should inspire people interested in applying digital twins for tighter cybersecurity. In addition to following the suggestions here, individuals should consider how other industry peers use these tools for better cyberattack protection.

How to Improve Cybersecurity With Digital Twins

ডিজিটাল টুইনস দিয়ে কীভাবে সাইবার সিকিউরিটি উন্নত করা যায়

So verbessern Sie die Cybersicherheit mit digitalen Zwillingen

Cómo mejorar la ciberseguridad con gemelos digitales

Comment améliorer la cybersécurité avec les jumeaux numériques

डिजिटल ट्विन्स के साथ साइबर सुरक्षा कैसे सुधारें

デジタルツインでサイバーセキュリティを向上させる方法

디지털 트윈으로 사이버 보안을 개선하는 방법

Como melhorar a segurança cibernética com gêmeos digitais

Как улучшить кибербезопасность с помощью цифровых двойников

Dijital İkizlerle Siber Güvenlik Nasıl Geliştirilir?

Cách cải thiện an ninh mạng với bản song sinh kỹ thuật số

如何利用数字孪生提高网络安全

This week, I am taking a slight detour from my usual articles that explain AI Concepts or coding tutorials. Yes, it's a philosophical, nostalgic tour down the road I was fortunate enough to tread all along. One of the rare privileges of starting in a field quite early is that you see the field grow like your baby. When I started my Masters in Computer Vision a decade ago, a CS graduate friend of mine asked, “What does computer vision even mean?” It's a trivial question today with enough videos of computer vision tasks like segmentation (see below) floating around the internet. A result of image segmentation — a Computer Vision task that was all over the internet when deep learning took off a decade ago. So, what has changed in the last decade? There are always two sides to a coin. Also, growth can be both good and bad. Closed Community — Small Was Beautiful They say that people in Iceland are somehow related to each other and if someone is beyond your second cousin, you are good to marry them. Likewise, the AI community was close-knit. Gone are those feelings of a tight-knit community. When interacting with anyone from research you generally knew which group they belonged to. You somehow would relate to and respect their work as you read and appreciated their impact on the field. As the field is blowing up endlessly, it's becoming increasingly difficult to even say that you know someone’s work, forget about them. There are too many new names and novel branches that it's even challenging to keep track of. As another example, PyTorch was in its infancy. The Slack community was so small and helpful that the developers of PyTorch directly responded to questions we had about using the library. This encouraged learning more and more about it. Today, the framework is so mature, and new frameworks like LangChain and Llamaindex have propped up. The focus is all on LLMs over any other branch of AI. Hardware Luxury Those days, we used to train a deep neural network on a single GPU to create something impactful. Most work published in premier conferences like CVPR, NeurIPS, and ICML could be trained and replicated on a single 8 GB GPU machine or in the worst case, a single machine with 4 GPUs in rare cases. An image of the latest GPU clusters used by Tesla for AI (source: https://www.hpcwire.com/off-the-wire/tesla-to-roll-out-advanced-supercomputer) I vividly remember how happy I was to buy a single commodity GPU with just 8 GB of RAM to participate in Kaggle competitions. Some of the winning solutions from Kaggle Grandmasters trained deep learning models on a single GPU machine at home. Today’s AI world needs a GPU cluster to train foundational models. Even fine-tuning these models needs 24GB GPUs which are beasts, expensive, and affordable only by businesses with an “AI budget.” Skillset Demand Those were days when the field did not yet get the AI wrapper around it. Recruiters were given this alien task of hunting for “deep learning” engineers. Recruiters and start-up founders were hunting for deep learning experts through all channels. It was a norm to receive regular messages on LinkedIn asking me to join their team as a deep learning engineer. A screenshot of a job advert for a Machine Learning Engineer posted only 5 days ago receiving over 100 applicants! The current situation is that openings for “Machine Learning Engineer” get over 100 applications on LinkedIn well within a day of advertising. Take a look at the screenshot above if you find it hard to believe. The question remains as to how relevant the applicant's skills are to the job spec. But the market is saturated with skills — quite quickly! Sub-skills — Operations and Architecture Growth means more diversity and opportunity. There are new roles like ML-ops, LLM-ops, and ML-architects cropping up. Gone are those days of tiny, single-model files (< 1 GB). The growth of models both in size and capabilities has given rise to new skills in deploying and maintaining them. Moreover, training and deployment of models are being automated with tools such as MLFLow. The cloud infrastructure for training needs to be sophisticated enough. All this has given rise to full-time roles with dedicated responsibilities. Bye Bye ML Engineer, Hello AI Engineer The most fun in working in AI is writing the model architecture, and training the model from scratch using our in-house data. Though this involves pre-processing the data a lot, training models, and visualizing the training results used to be a lot of fun. There was/still is a specialized role for this called Machine Learning (ML) Engineer. The development of foundation models from tech giants is redefining this role. As model sizes grow, training budgets are enormous. As a matter of fact, the cost of training the LLama 2 model was $20m for Meta. Clearly start-ups or organizations trying to adopt AI do not wish to throw away this sum of money. It is now established that foundation models are for the tech giants to develop, with the exception of some companies like Mistral and Anthropic. Sadly, this means the ML Engineer role is getting cast into an AI Engineer role. The ML Engineer role was all about developing model architectures, training, and evaluating them. The new AI Engineer role mostly involves developing APIs or invoking APIs provided by Tech giants (OpenAI, Meta, and Google) to prompt the foundation models. In rare cases, it involves fine-tuning these foundation models. But companies have the option to build RAG pipelines or use the foundation models “as-is” over fine-tuning them. Conclusion To conclude, I see this as a slow amalgamation of the Software engineering and Machine Learning roles. The line between software engineers and deep learning experts is fading. So, in the years to come, software engineers will be AI engineers who work alongside foundation models, both to write code and to solve customer needs. Furthermore, in the coming years, companies will be grouped into two categories — the AI product and AI services companies. AI product companies would again be OpenAI, Meta, and Google to develop foundational models. AI services companies would provide API services by fine-tuning or developing RAG-style pipelines around AI foundational models to serve their customers. Lastly, does the spike in job applications indicate an impending bursting of the bubble like the dot com bubble? I feel so, YES. But let’s wait and watch it… Shout Out Hope that was useful. If you liked this article, why not follow me on Twitter for research updates, video tutorial links, and new AI tools from top AI labs? Also, please subscribe to my YouTube channel where I explain AI concepts and papers visually. This week, I am taking a slight detour from my usual articles that explain AI Concepts or coding tutorials. Yes, it's a philosophical, nostalgic tour down the road I was fortunate enough to tread all along. One of the rare privileges of starting in a field quite early is that you see the field grow like your baby. When I started my Masters in Computer Vision a decade ago, a CS graduate friend of mine asked, “What does computer vision even mean?” It's a trivial question today with enough videos of computer vision tasks like segmentation (see below) floating around the internet. A result of image segmentation — a Computer Vision task that was all over the internet when deep learning took off a decade ago. So, what has changed in the last decade? There are always two sides to a coin. Also, growth can be both good and bad. Closed Community — Small Was Beautiful Closed Community — Small Was Beautiful They say that people in Iceland are somehow related to each other and if someone is beyond your second cousin, you are good to marry them. Likewise, the AI community was close-knit. Gone are those feelings of a tight-knit community. When interacting with anyone from research you generally knew which group they belonged to. You somehow would relate to and respect their work as you read and appreciated their impact on the field. As the field is blowing up endlessly, it's becoming increasingly difficult to even say that you know someone’s work, forget about them. There are too many new names and novel branches that it's even challenging to keep track of. As another example, PyTorch was in its infancy. The Slack community was so small and helpful that the developers of PyTorch directly responded to questions we had about using the library. This encouraged learning more and more about it. Today, the framework is so mature, and new frameworks like LangChain and Llamaindex have propped up. The focus is all on LLMs over any other branch of AI. Hardware Luxury Hardware Luxury Those days, we used to train a deep neural network on a single GPU to create something impactful. Most work published in premier conferences like CVPR, NeurIPS, and ICML could be trained and replicated on a single 8 GB GPU machine or in the worst case, a single machine with 4 GPUs in rare cases. An image of the latest GPU clusters used by Tesla for AI (source: https://www.hpcwire.com/off-the-wire/tesla-to-roll-out-advanced-supercomputer ) https://www.hpcwire.com/off-the-wire/tesla-to-roll-out-advanced-supercomputer https://www.hpcwire.com/off-the-wire/tesla-to-roll-out-advanced-supercomputer I vividly remember how happy I was to buy a single commodity GPU with just 8 GB of RAM to participate in Kaggle competitions. Some of the winning solutions from Kaggle Grandmasters trained deep learning models on a single GPU machine at home. Today’s AI world needs a GPU cluster to train foundational models. Even fine-tuning these models needs 24GB GPUs which are beasts, expensive, and affordable only by businesses with an “AI budget.” Skillset Demand Skillset Demand Those were days when the field did not yet get the AI wrapper around it. Recruiters were given this alien task of hunting for “deep learning” engineers. Recruiters and start-up founders were hunting for deep learning experts through all channels. It was a norm to receive regular messages on LinkedIn asking me to join their team as a deep learning engineer. A screenshot of a job advert for a Machine Learning Engineer posted only 5 days ago receiving over 100 applicants! The current situation is that openings for “Machine Learning Engineer” get over 100 applications on LinkedIn well within a day of advertising. Take a look at the screenshot above if you find it hard to believe. The question remains as to how relevant the applicant's skills are to the job spec. But the market is saturated with skills — quite quickly! Sub-skills — Operations and Architecture Growth means more diversity and opportunity. There are new roles like ML-ops, LLM-ops, and ML-architects cropping up. Gone are those days of tiny, single-model files (< 1 GB). The growth of models both in size and capabilities has given rise to new skills in deploying and maintaining them. Moreover, training and deployment of models are being automated with tools such as MLFLow. The cloud infrastructure for training needs to be sophisticated enough. All this has given rise to full-time roles with dedicated responsibilities. Bye Bye ML Engineer, Hello AI Engineer The most fun in working in AI is writing the model architecture, and training the model from scratch using our in-house data. Though this involves pre-processing the data a lot, training models, and visualizing the training results used to be a lot of fun. There was/still is a specialized role for this called Machine Learning (ML) Engineer. The development of foundation models from tech giants is redefining this role. As model sizes grow, training budgets are enormous. As a matter of fact, the cost of training the LLama 2 model was $20m for Meta. Clearly start-ups or organizations trying to adopt AI do not wish to throw away this sum of money. It is now established that foundation models are for the tech giants to develop, with the exception of some companies like Mistral and Anthropic. Sadly, this means the ML Engineer role is getting cast into an AI Engineer role. The ML Engineer role was all about developing model architectures, training, and evaluating them. The new AI Engineer role mostly involves developing APIs or invoking APIs provided by Tech giants (OpenAI, Meta, and Google) to prompt the foundation models. In rare cases, it involves fine-tuning these foundation models. But companies have the option to build RAG pipelines or use the foundation models “as-is” over fine-tuning them. Conclusion To conclude, I see this as a slow amalgamation of the Software engineering and Machine Learning roles. The line between software engineers and deep learning experts is fading. So, in the years to come, software engineers will be AI engineers who work alongside foundation models, both to write code and to solve customer needs. Furthermore, in the coming years, companies will be grouped into two categories — the AI product and AI services companies. AI product companies would again be OpenAI, Meta, and Google to develop foundational models. AI services companies would provide API services by fine-tuning or developing RAG-style pipelines around AI foundational models to serve their customers. Lastly, does the spike in job applications indicate an impending bursting of the bubble like the dot com bubble? I feel so, YES. But let’s wait and watch it… Shout Out Hope that was useful. If you liked this article, why not follow me on Twitter for research updates, video tutorial links, and new AI tools from top AI labs? If you liked this article, why not follow me on Twitter for research updates, video tutorial links, and new AI tools from top AI labs? Twitter Twitter Also, please subscribe to my YouTube channel where I explain AI concepts and papers visually. Also, please subscribe to my YouTube channel where I explain AI concepts and papers visually. YouTube channel YouTube channel

Stories, News & Mentions about SPIKE

Spike’s Stories on HackerNoon