Small Datum: Perf regressions in MySQL from 5.6.21 to 8.0.35 using sysbench and a medium server, part 3

Tuesday, October 31, 2023

This is part 3 of my results from sysbench vs MySQL on a medium server - see part 1 and part 2. This post has results for all MySQL 5.7 releases.

tl;dr: QPS for MySQL 5.7.43 relative to 5.7.10

For point queries is similar
For range queries is ~8% slower except for full table scans where it is ~21% slower
For writes is up to 20% better

Results

The spreadsheet with all of the numbers is here.

For all charts below:

The y-axis doesn't start at 0 to improve readability.
Each section has four charts. The first two charts have many data points and aren't as easy to read as the third and fourth charts:

The charts show relative throughput which is (QPS for $me / QPS for MySQL 5.7.10). When that value is less than one then the version is slower than MySQL 5.7.10.

Point queries, part 1

Summary

QPS for MySQL 5.7.43 is similar to 5.7.10 on the medium server. There were big improvements early in 5.7 and there have been gradual regressions since then.
QPS for MySQL 5.7.43 is about 5% less than 5.7.10 on the small server

Point queries, part 2

Summary

QPS for MySQL 5.7.43 is similar to 5.7.10 on the medium server. There were big improvements early in 5.7 and there have been gradual regressions since then.
QPS for MySQL 5.7.43 is about 5% less than 5.7.10 on the small server

Range queries, part 1

Summary

QPS for MySQL 5.7.43 is about 8% less than 5.7.10 on the medium server for most microbenchmarks
QPS for MySQL 5.7.43 on full table scans (scan_range=...) is about 21% less than 5.7.10 on the medium server

Range queries, part 2

Summary

QPS for MySQL 5.7.43 is similar to 5.7.10 on the medium server for most microbenchmarks
QPS for MySQL 5.7.43 is about 8% less than 5.7.10 when the range scan length is large