New Tool for Benchmarking EVM Nodes Over RPC
Learn about a new tool for benchmarking EVM nodes over RPC. Find out how Flood can help optimize RPC performance, and get tips on generating success, throughput, and latency graphs.

Georgios Konstantopoulos
very greek accent. cto & research partner @paradigm. optimist and rustacean 🦀 // we're hiring - dm if you wanna work together on hard internal eng problems

-
NEW tool @paradigm Github!
— Georgios Konstantopoulos (@gakonst) June 6, 2023
This time @notnotstorm publishes Flood, a load-testing tool for benchmarking EVM nodes over RPC.
We've been using Flood to optimize Reth's RPC performance (public alpha soon!), and we're excited to share it with the eco.https://t.co/7ZCKSZdEmu 1/n https://t.co/UqXSE1Cd7n -
1 month ago I asked for a tool for this job, but got no seemingly great answers. so we built it.https://t.co/ZGc61DGe2D 2/n
— Georgios Konstantopoulos (@gakonst) June 6, 2023 -
For each RPC call, Flood generates 3 graphs:
— Georgios Konstantopoulos (@gakonst) June 6, 2023
1. Success vs Request Rate
2. Throughput vs Request Rate
3. Latency vs Success Rate
Here's an example report:https://t.co/yZec3rdSVq 3/n pic.twitter.com/LdlqGzbknF -
Requests are sampled using data from the Paradigm Data Portal: https://t.co/WZhVV70AwH. This contains Parquet files of historical onchain data.
— Georgios Konstantopoulos (@gakonst) June 6, 2023
We query that to sample a realistic dataset with production workloads for our benchmarks.
(also built by @notnotstorm!) 4/n -
Kudos to @achalvs for the kickass graphics also. 🌊🌊🌊 5/n pic.twitter.com/wCA6lsM0Oo
— Georgios Konstantopoulos (@gakonst) June 6, 2023 -
With Flood, we seek to shed light on the performance of RPC providers/nodes under heavy load, create useful benchmarks to close feedback loops & ultimately let customers choose the best solutions.
— Georgios Konstantopoulos (@gakonst) June 6, 2023
See you on Github! 6/6 fin!https://t.co/7ZCKSZdEmu -
Read the full blog:https://t.co/rNqWMIWoWt
— Georgios Konstantopoulos (@gakonst) June 6, 2023