Django MariaDB-Galera Perfermance on K8s Django MariaDB-Galera Perfermance on K8s kubernetes kubernetes

Django MariaDB-Galera Perfermance on K8s


INSERT one row at a time is slow. These are faster:

  • Batched insert (many rows in a single INSERT). Recommend 100 to 1000 rows at a time.
  • LOAD DATA INFILE

The Query Cache slows down inserts. Anyway, it is not allowed on Galera.

How far apart (in milliseconds) are the nodes?

These are terribly high, go back to the defaults:

tmp_table_size=2Gmax_heap_table_size=2G

Some places specify utf8, some specify utf8mb4. The latter is preferred. The inconsistency may be causing some performance lag.