stephenyates-gc commented 3 years ago

Description

As an ops lead, I want minimize size of data on wire and speed up repetitive retrieval of data And minimize load on the database, So that I have a faster system overall.

WHY are we building?

Fasten our system, reduce hit on the database.

In the context of the 2023-06-01-notifications-getting-errors-repeatedly incident, we got payloads that were larger than expected and caused issues, having a cache system would help offload the database.

WHAT are we building?

A cache layer between the app and the database, potentially leveraging SQL alchemy plugin such as Dogpile caching.

VALUE created by our solution

Faster system, more throughput.

Notes

Acceptance Criteria

[ ] start small with minimal amount of usable caching (one table? one column?)
[ ] measure before and after to see improvement
[ ] System behaves the same as before
[ ] Option to flush the cache easily when necessary (hook into existing cache flushing perhaps).

QA Steps

[ ] Run performance testing with and without the caching support, measure outcome.

yaelberger-commits commented 2 years ago

@jimleroyer Is this card still relevant or has it already been completed? Thanks!

jimleroyer commented 2 years ago

@yaelberger-commits Still relevant and still in the talks. That would ease up bottlenecks on the database if we are having troubles.

cds-snc / notification-planning-core

Add/Rework caching system between API/admin and database #8