Layer-2 solutions are seen as the holy grail of scaling the Ethereum Blockchain, and it is evident why that is the case.
Side chains such as Polygon, Ronin, or Gnosis permit users to interact with the Ethereum eco-system at a low cost, lifting the eco-system throughput to over 100 transactions per second.
So, it is no surprise that great things are expected from layer-2 solutions to finally push Ethereum on visa-like throughput levels.
However, while some Layer-2 solutions benefit from Ethereum main chain security guarantees, a significant portion of the Layer-2 throughput stems from side chains that almost solely rely on their own security model which is significantly weaker than the one Ethereum provides.
This deterioration of the security guarantees is not apparent to the end users and can come at a significant cost.
Examples of this are the Ronin Bridge Exploit and the Harmony Horizon Bridge Hack which were already discussed in a past article here: https://hackernoon.com/the-biggest-recent-cryptocurrency-hacks-what-is-behind-them
Luckily, there are Layer-2 solutions that offer identical safety guarantees as the main chain. The most prominent group of these solutions is Rollups.
Vitalik described Rollups in detail in https://vitalik.ca/general/2021/01/05/rollup.html, already estimating the scalability potential of 1000-4000 transactions per second.
While he mentions that user behavior does play a role, he still expects performance gains in the order of x100 compared to the main chain performance.
Unfortunately, Reality Looks Different
A recently published study that analyzed layer-2 solutions over the course of 2021 shows that given the usage patterns, Rollups stray far from the expected scalability goals and stay way below 200 transactions per second and even without deposits and withdrawals below 300.
Based on the obtained usage patterns, even Side-Chains such as Polygon and Ronin struggle to reach high throughput levels due to the resource consumption from deposits and withdrawals.
While this improved significantly after receiving support from centralized exchanges, this just further deteriorates the security guarantees.
While progress on cross-layer rollups can improve the performance losses from on and off-boarding, as shown in the study, the base cost of Rollups is still too high to reach visa-level scalability goals.
Now, What Can We Learn From This?
First, users should avoid using Layer-2 solutions with questionable security models.
Second, while Rollups can help with scalability, more emphasis has to be put on main-chain performance improvements to fully leverage the potential of Layer-2 Solutions.