Workaround · Performance · Annoyance
Recursive CTEs, cross-database joins and large counts needed workarounds
1 source thread · first seen 2022-04
Summary
On a large mature Postgres database Hasura scaled well, but recursive queries, joins across databases and very large counts required workarounds.
- Affects
- team running Hasura over a 1 TB+ database with thousands of concurrent users
- Workaround
- Approximate counts via HLL or planner estimates
Evidence
Excerpts are copied word for word from the source; follow the link to read it in full.
“Some areas that needed a bit work around for me - recursive CTEs, cross-db joins, performance of very large counts”