Log Axis — Throughput Across Four Orders of Magnitude
Measured throughput against request size on a logarithmic y axis, with the axis stating its own base. A log axis is the right choice here and the wrong choice more often than it is used, so the template pairs it with a note saying what the reader has to do differently: on this axis equal vertical distances are equal RATIOS, not equal differences.
Make it your own.
title: Sustained throughput by request size
subtitle: Single node, warm cache, median of five runs
x: Request size (KiB)
y: Throughput (requests/s)
x scale: log
y scale: log
grid: both
legend: right
line "v2.4" curve: monotone
1 48200
4 31500
16 14800
64 4900
256 1310
1024 338
4096 84
line "v2.5" curve: monotone
1 51900
4 36200
16 18400
64 6700
256 1980
1024 552
4096 147
note: Both axes are logarithmic. Equal vertical distances on this figure are equal RATIOS, not equal differences — the gap between the two curves is a constant factor of about 1.5, not a constant number of requests.