An Off-The-Chain Execution Environment for Scalable Testing and Profiling...
summary
An off‑chain replayer stores per-transaction input-output tuples in compact substates and replays 590 million historical transactions from the first 9M blocks with 100% replay accuracy, shrinking replay storage from 700.11 GB to 285.39 GB and boosting throughput from 620.62 tx/s to 2,817.98 tx/s single-threaded and 30,168.76 tx/s on 44 cores (full replay finishes in ~6 hours). The system powers parallel fuzzing, a wasteful-instruction metric (75 hours for 9M blocks), and hard‑fork regression testing.
AI-written summary · about
chapters (17)
- 0:00Intro
- 0:22Problem: Transaction Replay
- 1:17Transaction Replay on Geth Archive Node
- 1:57Transaction Replay on Geth Full Node
- 2:44Efficient Transaction Replay is an Unsolved Problem
- 3:20Contributions
- 4:05Solution (Overview)
- 5:13Off-The-Chain Testing - Recorder
- 6:39Off-The-Chain Testing - Replayer
- 7:58Evaluation - System Configuration
- 8:17Evaluation - Replay Performance (Space)
- 8:27Evaluation - Replay Performance (Time)
- 8:52Use Cases
- 9:35Metric (cont.)
- 10:24Fuzzer (cont.)
- 11:24Hard Fork Assessment (cont.)
- 12:21Conclusion
related talks
33mContainerized Time Travel: Replicating Production Performance
01h 12mScaling distributed ledgers: Achieving arbitrary throughput and sub-second latency
25mOperation-level Concurrent Transaction Execution for Ethereum
50mTrusted Execution Environments Meet the Blockchain
18mVegeta: Enabling Parallel Smart Contract Execution in Leaderless Blockchains
24mValidity and the future of blockchains
42mReplay Production Traffic Locally: Instantly Spot & Fix Breaking Changes
50mDistributed Replay Testing with Your Data, Your Way!