Mehdi Akiki

Posts tagged "statistics"

  • Published on

    Investigation · Part 9 of 20 · When the benchmark lies

    Coordinated Omission: How a Load Test Hides Its Worst Latency

    My toy server stops for half a second every five seconds. A client that waits for each answer saw 4 slow requests out of 1.36 million, and a p99 of 30 microseconds. A client with a fixed schedule saw 1,608 slow requests and a p99 of 450 milliseconds.
  • Published on

    Investigation · Part 8 of 20 · When the benchmark lies

    The Average Hides the Slow Requests

    I timed four million single Vec::push calls: mean 22.8 ns, slowest 315,684 ns. The slow ones were not the reallocations but page faults, and huge pages made them bigger. Then a simulated request handler where the mean is a time that no request had.
  • Published on

    Investigation · Part 7 of 10 · Inside a database index

    Planner Statistics Go Stale Too: ANALYZE and a Plan That Changes

    Without statistics, SQLite picked the wrong index and took 97 ms instead of 0.9 ms. ANALYZE fixed it. Then I changed the data, and the old statistics kept a slow plan that PRAGMA optimize did not see. Statistics are derived data too, and they can also live in an open connection.