Select statements that are in execution during database CPU utilization peak times - namespaces table

After investigating spikes over 70% of CPU usage on the database here:

Link to reports/investigations

we find out this statement being executed thousands of times:

Query
SELECT "namespaces".* FROM "namespaces" WHERE "namespaces"."id" = $1 LIMIT $2

QueryId: -765625686214592666 6974950735891200787

This query is being executed from 500 to 1000 times a second:

Could be possible creating or improving our caching on those tables? Could we have a unified cache to query, and when we execute updates on it, we could refresh the cache?

Possible places in application

worker class	count (query?)	notes
ProjectImportScheduleWorker	6
ExpireJobCacheWorker	6	Addressed with !57773 (merged)
RepositoryUpdateMirrorWorker	5
Namespaces::ScheduleAggregationWorker	4
BuildHooksWorker	4
PipelineProcessWorker	3	Awaiting ci_remove_update_retried_from_process_pipeline to be enabled
ExpirePipelineCacheWorker	3	!57304 (merged)
ProcessCommitWorker	2
Notify	2
PostReceive	2
RequirementsManagement::ProcessRequirementsReportsWorker	1
Deployments::ExecuteHooksWorker	1
Deployments::UpdateEnvironmentWorker	1
MergeRequestMergeabilityCheckWorker	1
BuildFinishedWorker	1

source: https://gitlab.com/gitlab-com/gl-infra/infrastructure/-/issues/12361#note_487435760

json.class.keyword: Descending	Job Count last 24 hours
ExpirePipelineCacheWorker	10220262
PipelineProcessWorker	8226159
ExpireJobCacheWorker	7414267
BuildHooksWorker	7354278
Namespaces::ScheduleAggregationWorker	4659909
ProjectImportScheduleWorker	3149266
RepositoryUpdateMirrorWorker	3120667
BuildFinishedWorker	1927207
RequirementsManagement::ProcessRequirementsReportsWorker	1924421
PostReceive	1418632
ProcessCommitWorker	948680
Deployments::ExecuteHooksWorker	366663
MergeRequestMergeabilityCheckWorker	265358
Deployments::UpdateEnvironmentWorker	149465
ProcessCommitWorker	948,680
Deployments::ExecuteHooksWorker	366,663
MergeRequestMergeabilityCheckWorker	265,358
Deployments::UpdateEnvironmentWorker	149,465

source: https://log.gprd.gitlab.net/goto/264cebbbd9426bddae9ee512a0cc8a6e

Metrics

Collected from https://gitlab.com/gitlab-com/gl-infra/infrastructure/-/snippets/2047044

Metric	Measurement
Total number of calls
% of Total time
TPS
Duration