benchmarking – Page 6 – Türker Yıldırım

An index that was never used, and the week it took to prove it

Deep diveNovember 8, 201910 min read

Forty indexes on one table and nobody willing to drop any of them. Invisible indexes turn a guess into a week of production traffic as evidence.

The N+1 in the ORM you cannot see from the template

Deep diveOctober 17, 201910 min read

Subquery selects in 6.0, and the aggregate that used to be a loop. Eager loading is all-or-nothing per relation, which is the problem.

A flame graph’s x axis is not time

NoteSeptember 19, 20191 min read

The instinct is to read it left to right as a timeline, and the x axis is alphabetical — a wide block is total time across the whole…

Measure the p99 and the max; the average hides incidents

NoteAugust 22, 20192 min read

An hourly spike that made every request four hundred milliseconds slower for two seconds moved the average by nine milliseconds and was invisible in every graph anyone looks…

The queue worker that leaked, and the memory limit that hid it

Deep diveJuly 15, 201910 min read

A worker restarting every eleven minutes, reported as healthy. Finding the leak is not the same as bounding it, and the seatbelt hides the next one.

Load testing with think time, or the number is fiction

NoteJuly 11, 20191 min read

A tool firing requests as fast as it can with no pause measures how quickly the server can be saturated, which corresponds to no real number of users.…

Lazy collections and the report that would not fit in memory

Deep diveJune 24, 201910 min read

A memory limit hit on the fourth of the month, every month. A collection is an array, and an array is all of it at once — until…

A cache that survives the thing it is caching

Deep diveMay 8, 201914 min read

The cache expired at noon and the database fell over at 12:00:01. A lock, a two-tier expiry, jittered TTLs, and versioned keys — in that order of importance.

Product lookup tables, and reading a shop without postmeta

Deep diveApril 10, 201910 min read

3.6 added lookup tables, which is WooCommerce admitting postmeta does not scale. What they contain, what they do not, and the regeneration nobody schedules.

The N+1 that does not appear in the query log

Deep diveJune 14, 201810 min read

340 queries on a page showing twenty rows, and none of them are slow. Every one is under the threshold, which is why the slow log has nothing…