Ethereum Status Optimization Experiment Achieves Breakthrough: Database Shrinks by 77.5%

An performance experiment conducted on the Ethereum status expiration mechanism has revealed impressive results. This study involved simulating the main Ethereum network load over a full year using the Geth client, comparing performance between nodes that store the entire history of status and nodes that only maintain active status data from the past year.

Experiment Design: Node Efficiency Testing Strategy

Ethereum execution layer researcher, weiihann, designed this experiment to test the practical impact of reducing status load. The experimental methodology focused on comparing two operational models: a conventional model that maintains full status since genesis, and an optimized model that only stores active status accessed within a one-year period. Data was collected through replaying real transaction loads from mainnet, ensuring the test scenarios reflected actual operational conditions.

Drastic Reduction: From 359GB to 81GB

The most striking result of this experiment was the reduction in database size. Nodes that only maintained one year’s status successfully reduced storage capacity from 359 GB to just 81 GB—a spectacular decrease of 77.5%. The largest reduction occurred in Trie structure storage, which is a critical component in storing Ethereum state data. This database optimization has significant implications: reducing hardware requirements for nodes, lowering entry barriers for node operators, and simultaneously opening space to increase gas limits and network throughput.

Performance Leap: Faster Execution and Lower Latency

Beyond database size, the experiment revealed dramatic improvements in execution performance. Block reprocessing time decreased by about 15%, indicating higher efficiency in re-executing historical loads. Latency metrics showed even more impressive improvements: P50 storage read latency decreased by 46%, while P99 latency dropped by 36%. These reductions are crucial as they affect the node’s experience in processing new transactions. Additionally, tail latency increased consistently, with P99 block insertion times reduced by 21%, helping nodes maintain synchronization with the network even under high load.

Next Steps: Exploring Various Expiration Scenarios

This research opens the door for further investigation. The next phase will compare results with other Ethereum clients, test alternative expiration cycles such as six months, and explore strategies focused on cleaning up storage of contracts that are no longer in use. Ongoing experiments demonstrate the Ethereum community’s commitment to long-term network scalability and efficiency.