At MigaLabs we are glad to release the report of our latest collaboration with the Obol team. Earlier this year we did a performance study on the Obol Distributed Validator Technology (DVT), running validators in different machines, cloud providers, and geographical locations.
Since September 2022, the Ethereum blockchain is composed of two layers: the Consensus Layer (CL) and the Execution Layer (EL). The Consensus Layer is formed by validators (a single node running 24/7 with a validated key), who decide which is the Ethereum canonical chain.
With DVT, a validator can be split into several nodes, removing the single-point-of-failure for validators, and creating an active-active redundancy with a failure threshold (depending on the cluster). This technology adds an extra layer of complexity to the validator mechanism and therefore it is important to study its performance.
The goal of the study was to compare the consensus performance of the Distributed Validators (DVs) vs conventional validators in the Ethereum ecosystem. As mentioned, the DVT adds some extra complexity, which could potentially induce higher latencies, missed duties, and fewer rewards.
To do this study, we measured the achieved consensus duties and their rewards, which define the performance of a validator.
The study involved:
21 different machines (from two different cloud providers and 4 continents)
3 different clusters (of sizes 4, 7, and 10 nodes)
3000 validators on the Goerli network
More than 10k epochs of measurements
During the study, we measured:
Hardware resource consumption
The latency between the nodes in the same cluster
Consensus duties achievement