PERFORMANCE Signal 398
Just One Function, 10x Faster? Reading a Rust Performance PR
↗
A single-function rewrite in GreptimeDB’s Prometheus remote-read path cuts the RecordBatch-to-TimeSeries conversion time by more than tenfold.
The conversion step previously consumed over a third of CPU in a realistic workload, limiting read-throughput. By integrating the optimized version, services can handle many more series per second without additional hardware. Existing deployments need only upgrade to the beta release that includes the change.
Written by elseif from the cluster below · every claim links back to a sourceThe three things worth knowing
The PR replaces the original recordbatches_to_timeseries implementation in src/servers/src/prom_store.rs.
Benchmarks show throughput rising from roughly 2 M rows / s to well over 20 M rows / s for typical series-by-samples patterns.
Dictionary-encoded label data remains slower than plain UTF-8 strings, indicating a separate optimization opportunity.
THE CLUSTER