Validated, staking on eth2: #5 - Why client diversity matters

The Power of Client Diversity in Eth2: Why It Matters

As the Ethereum 2.0 (Eth2) testnet, Medalla, continues to evolve, a crucial aspect of its design has come to the forefront: client diversity. With multiple clients available, each with its own strengths and weaknesses, the network's resilience and security are put to the test. In this article, we'll delve into the importance of client diversity in Eth2 and explore the implications of a dominant client on the network's stability.

The Medalla Testnet: A Wake-Up Call

When Medalla launched, users were encouraged to experiment with different clients. However, the results were far from ideal. Nimbus and Lodestar nodes struggled to cope with the workload of a full testnet, getting stuck and causing Medalla to fail to finalize for the first half hour of its existence. This was a stark reminder of the importance of client diversity and the potential risks of a dominant client.

The Prysm Time-Bug: A Case Study

Fast forward to August 14th, when Prysm nodes lost track of time due to a faulty timeserver. This led to a chain of events that highlighted the dangers of a dominant client. With Prysm nodes making up over 60% of the network, the threshold for finalizing blocks couldn't be met. The nodes then flooded the network with short forks, causing chaos and leading to nodes falling behind, needing to sync, and running out of memory.

The Incentives of Decentralization

In Eth2, decentralization is not just a buzzword; it's a deliberate design choice. The protocol is designed to incentivize validators to make decisions that are in the best interest of the network, rather than simply following a dominant client. The inactivity penalty, for example, kicks in when more than 1/3 of validators are offline, causing penalties to ramp up. This means that validators want to ensure that if something takes their node offline, it's unlikely to take many other nodes offline at the same time.

The Anti-Correlation Mechanisms

Eth2's design includes anti-correlation mechanisms that tie individual penalties to how much each validator is impacting the network. This means that if many validators are slashed at the same time as you, penalties go up to as high as 32 ETH. The point at which this happens is again the magic 1/3 threshold. This is a deliberate design choice to incentivize validators to make decisions that are in the best interest of the network.

The Numbers: A Dominant Client's Impact

If only 20% of Eth2 nodes were running Prysm and 85% of people were online, the inactivity penalty wouldn't have kicked in for Prysm nodes, and the problem could have been fixed with only minor penalties and some sleepless nights for the devs. However, because so many people were running the same client, somewhere between 3500 and 5000 validators were slashed in a short period of time. The high degree of correlation means that slashings were ~16 ETH for these validators because they were using a popular client.

Try Something New

Now is the time to experiment with different clients. Find a client that a minority of validators are using, and give it a try. Lighthouse, Teku, Nimbus, and Prysm are all reasonably stable at the moment while Lodestar is catching up fast. Most importantly, TRY A NEW CLIENT! We have an opportunity to create a more healthy distribution on Medalla in preparation for a decentralized mainnet.

Conclusion

Client diversity is a crucial aspect of Eth2's design, and it's essential to understand the implications of a dominant client on the network's stability. By experimenting with different clients and creating a more healthy distribution, we can ensure that Eth2 is a resilient and secure network. The incentives of decentralization and anti-correlation mechanisms are designed to encourage validators to make decisions that are in the best interest of the network. As we move forward, it's essential to prioritize client diversity and create a more decentralized and secure Eth2 network.

Source: https://blog.ethereum.org/en/2020/08/21/validated-why-client-diversity-matters