Gravity Intern

As part of the community that has followed Reth’s development since the early days, the @GravityChain team has achieved a new milestone: the integration of RocksDB, which delivers a significant performance leap for Reth. RocksDB is an ultra fast local storage engine designed for applications that require high throughput. github.com/facebook/rocksdb With a fully featured RocksDB backend including idempotency and crash recovery Gravity Reth recorded up to a 10× throughput improvement on a 30-million-state workload. This is not a small number. Baseline Reth: persistence every 1 second Gravity Reth + RocksDB: 250 ms That means 4x faster flush frequency and far more stable throughput. All benchmarks were performed on standard L1 hardware: 32 cores, 64 GB RAM, 350 MB/s SSD, 30k IOPS. Test state: 10 million accounts + 20 ERC20 entries RocksDB is not only fast, but also highly predictable, with a stable I/O profile and drastically reduced latency. This integration extends Gravity previously merged contributions to Reth, including: • Parallel transaction validation • Reduced tx pool • RPC correctness fixes • Gravity Reth achieving 1.5 – 1.9 gigagas/s As someone who has been following this project closely, I see RocksDB as a crucial foundation for the future of L1/L2: massive state, parallel execution, and consistently high throughput. The RocksDB backend is currently in the process of being upstreamed to Reth. github.com/paradigmxyz/reth/… @Quazorg @GravityIntern