Shardus

The foundation for global-scale decentralized applications.

The Shardus Consensus Algorithm: A Blockless Architecture for Immediate Transaction Processing

With the introduction of the Bitcoin network software, a new era of technology was born. It provided, for the first time, a trustless, peer-to-peer payment network with a very transparent, and rule-based money supply. It revealed a distributed ledger technology that will profoundly change the way society transfers value, and ushered in a new era of trusted decentralization.
Yet it soon became clear Bitcoin left considerable room for experimentation. Self-imposed limits on block size led many academics and engineers to explore alternatives in the form of forks and second-layer applications. Experimentation has now spawned over 2,000 start-up projects.
Litecoin tried to increase transaction speeds and become less centralized. Ethereum added a Turing-complete smart contract layer. Peercoin and Nxt looked to be more energy efficient. Dash and Monero focused on true anonymity.
Still, the first generation of improvements didn’t focus much on scalability. As the Bitcoin community grew and the number of transactions increased, the scaling issue became a major concern. Fundamental bottlenecks in Bitcoin limit the ability of its current peer-to-peer overlay network to support substantially higher throughputs and lower latencies. This has led to a second generation of improvements, and an explosion of new token-based start-ups.
Bitshares, Ripple and Stellar were the first to increase transaction speeds above 1000/sec. Each of these second generation efforts took a unique approach to solving the global scalability issue.
A third generation of distributed ledger technology has emerged, one which sees innovation in blockless data structures. Nano and IoTA use Directed Acyclic Graphs (DAG) to significantly improve transaction throughput compared to blockchain architectures. Hashgraph uses a different blockless architecture to achieve significant throughput.
Still, current distributed ledger protocols remain vertically scalable, but do not scale horizontally. In each case, adding more nodes to the network does not increase throughput or capacity. The scalability issue remains unsolved, and major advances and adoption will require a basic rethinking of technical approaches.
Introducing the Shardus Ledger
The Shardus Ledger is a fourth generation distributed ledger that utilizes a blockless architecture. By processing each transaction separately, instead of into blocks, we are able to achieve true scalability and dramatically improve throughput.
This is achieved through a decentralized ledger that is sharded to evenly distribute compute, storage and bandwidth across all nodes. It includes a unique consensus protocol, The Shardus Consensus Algorithm, to support immediate processing of transactions without grouping them into blocks by a leader or a temporary leader node.
New Code Base — not forked from any other network
The Shardus Consensus Algorithm — unique consensus model
The Shardus Ledger — adding more nodes increases compute and storage capacity of the network while bandwidth remains constant
The Shardus Token — incentive token for development
Liberdus — the first application, a coin and peer-to-peer payment network, will feature Shardus technology
Our motivation to create the Shardus technology is to support global scale decentralized applications which aim to reach billions of users and require millions of transactions per second.
The Shardus community strives to build a network that is:
Horizontally Scalable — increases both throughput and capacity
Bandwidth Constant — transaction bandwidth does not increase by
number of nodes
Auto-scaling — self-governing in number of nodes necessary to
maintain network
Low-latency — time to confirmation the network has applied the transaction faster than blocks
Faster Finality — time to confirmation the transaction irreversible faster
than blocks
Fair — transactions are processed and applied in order received
Secure — Byzantine fault tolerant, cryptographically secure
Sustainable — supported by a native development fund to maintain and
promote the network
Self-governed — all coin holders can participate in the future direction
of the network
Mining and Liberdus
In Bitcoin-like networks, miners tend to have the most control because they choose which version of the software to run. In addition, the energy required to confirm Bitcoin transactions will soon exceed 1% of the world’s energy. While Ethereum has discussed reducing energy through Proof of Stake, they have yet to achieve it.
Instead of miners competing through Proof of Work for the native Liberdus Coin, Shardus introduces a unique consensus algorithm in which steady incentives are given to nodes on a regular basis. Inspired by the concept of universal basic income, it rewards nodes daily for providing resources to the network. Nodes are thus highly incentivized to self-govern the network.
Decentralization
One simple measure of network centralization is the largest percentage of resources provided by a single entity or group of related entities. For truly decentralized systems, nodes providing resources to the network must be unrelated. Ideally, each node would originate from an independent entity and would be unknown to any other. This was a key tenant of Satoshi’s peer-to-peer trustless cash system.
Yet Bitcoin experiences centralization challenges in its mining model, one that puts control in the hands of large mining farms and those with cheap energy resources. EOS uses a group of designated block producers determined by the community, relinquishing decentralization. Other efforts describe private blockchain structures in which some level of centralization is a virtue. Yet many of the security properties of decentralization fail when a network achieves centralization of 33% or more.
Shardus aims for a high level of decentralization by providing steady incentives to nodes, and auto-scaling to prevent large numbers of nodes joining at once. The Shardus Ledger ensures no single node or group of nodes can control the network.
No Fees Necessary
Transactions on the Shardus Ledger do not require a small payment for a transaction to be processed faster. In blockchain-based networks, the maximum size of the block creates a scarcity of space, making transaction fees unavoidable.
Fees are also used to as an anti-spam/DOS measure, making costs of bad behavior higher than the actual benefits. Anyone attempting to congest the network by broadcasting many invalid transactions must pay more in fees than legitimate users.
Networks often require a minimum transaction fee if the amount being transferred goes below some threshold. This discourages micro transactions and slows global adoption, making the minimum fee potentially larger than the transaction amount.
Shardus will replace transaction fees with variable, just-in-time proof of work. No fees are required to complete transactions, however the community may vote on models including transaction fees which are burned in order to reduce the supply of coins.
Security is Key
The level of security in a trusted environment is directly proportional to its level of centralization. When the level of centralization increases beyond some threshold, the network will operate outside the bounds of common operational models and security can no longer be guaranteed. Typical worst-case scenarios include a network partition, a malicious change to state data, or denial of service attacks.
Shardus assumes four operational models: honest majority, uncoordinated majority, coordinated choice and bribery. The project aims to provide the highest level of security possible under common operational models, avoiding high levels of centralization in our consensus and network protocol.
Sustainability
Shardus will include a native development fund to ensure future development and maintenance of the network software. Bitcoin provides funds only for miners — for nodes validating transactions — but core developers are supported by outside contributions and the Bitcoin Foundation. Many projects have begun to see the need for future planning, however. The Dash network was one of the first to establish a native fund to support future development and promotion of the network.
Self-governance
In Bitcoin-like networks, a core group of developers controls changes to the software. Its governance model for changes and upgrades is one controlled by a relatively small group. Recent developments, however, include User-activated Soft Forks (UASFs), in which an economic majority can activate a change in the software without the involvement of miners.
Shardus aims to incorporate features similar to UASFs to allow the entire community of coin holders to participate in the future direction of the network.
Sharding to Achieve Global Scale
Shardus will achieve global scalability through sharding at the state, compute and storage levels. Sharding, a horizontal scaling method for data, has existed for decades and is being explored, mostly in retrospect, by some blockchain efforts such as EOS, Zilliqa, Radix and Hashgraph. Ethereum plans to introduce sharding by 2020, but only to break the compute bottleneck.
In the Shardus model, an increase in the number of participating nodes increases compute and storage capacity of the network while keeping bandwidth requirements constant. The ledger is sharded evenly across the network, reducing the load on any given node or groups. The UBL platform also allows for side chains, Lightning Network-style solutions, and smart contracts to be incorporated for additional scaling of other distributed ledgers.
No Initial Coin Offering
The Shardus model also takes a unique approach to incentives. Instead of expending effort on marketing an ICO, the project incentivizes early developers with an ERC20 token to participate in the network’s creation, then provides a conversion method once the main net goes live. Once the network is created — reasonably and transparently rewarding co-founders and early contributors — the code will be open source through a Creative Commons license.
Join Shardus
Node.js, Go, or Haskell developers are encouraged to sign on to the project via a series of qualifying exercises. There is a bounty for recruiting developers to the program, as well as bounty projects for helping us raise awareness via Twitter, Telegram, Medium and third-party media. Check Github or our website for details.

Tags

Topics of interest