Optimize project counters with repositories enabled counter

What does this MR do?

Optimize query produced by https://gitlab.com/gitlab-org/gitlab/-/blob/908902d1d4f6fbcd7780150c780837fd4fa8b301/ee/lib/ee/gitlab/usage_data.rb#L241

projects_with_repositories_enabled: distinct_count(::Project.with_repositories_enabled.where(time_period), :creator_id),

query with time constraint and batching

SELECT
    COUNT(DISTINCT "projects"."creator_id")
FROM
    "projects"
    INNER JOIN "project_features" ON "project_features"."project_id" = "projects"."id"
WHERE
    "project_features"."repository_access_level" = 20
    AND "projects"."created_at" BETWEEN '2020-02-07 21:17:04.618237'
    AND '2020-03-06 21:17:04.618286'
    AND "projects"."creator_id" >= 810000
    AND "projects"."creator_id" < 820000;

query with no time constraint and batching

SELECT
    COUNT(DISTINCT "projects"."creator_id")
FROM
    "projects"
    INNER JOIN "project_features" ON "project_features"."project_id" = "projects"."id"
WHERE
    "project_features"."repository_access_level" = 20
    AND "projects"."creator_id" >= 810000
    AND "projects"."creator_id" < 820000;

explain data from database-lab

• id between 1 AND 10_000 and creator_id between 1 AND 100_000

Before

• with time constraint - https://explain.depesz.com/s/V5dT
• with no time constraint - https://explain.depesz.com/s/e4fO

Observations

1. filter applied when join performed between projects and project_features for the project_id and repository_access_level

Index Cond: (project_features.project_id = projects.id)
Filter: (project_features.repository_access_level = 20)

• combat this with adding index

CREATE INDEX index_project_features_on_project_id_and_repository_access_level_20
ON project_features(project_id)
where project_features.repository_access_level = 20;

• explain now shows index only condition hit(improvement) - https://explain.depesz.com/s/obCh

Index Cond: (project_features.project_id = projects.id)

2. MAX/MIN calculations

• Query - SELECT MAX("projects"."creator_id") FROM "projects" WHERE "projects"."created_at" BETWEEN '2020-02-07 06:22:09.650617' AND '2020-03-06 06:22:09.650886'
• Explain results(no index, but within thresholds) - https://explain.depesz.com/s/SDPC

Plan

• Add this indexes

CREATE INDEX index_project_features_on_project_id_and_repository_access_level_20
ON project_features(project_id, repository_access_level)
where project_features.repository_access_level = 20;

After only index conditions are hit

• with time constraint - https://explain.depesz.com/s/kD9P
• with no time constraint - https://explain.depesz.com/s/obCh

Timing

After the index for batch counting takes 55 seconds pessimistic

• 5.5 million users,
• with 1_250 batch sizes
• 5.5M/1_250 = 4_400 loops
• Time: < 1.4s ( cold cache )

Migration output

10:10 $ VERBOSE=true be rake db:migrate:up VERSION=20200309140540
== 20200309140540 AddIndexOnProjectIdAndRepositoryAccessLevelToProjectFeatures: migrating
-- transaction_open?()
   -> 0.0000s
-- index_exists?(:project_features, :project_id, {:where=>"repository_access_level = 20", :name=>"index_project_features_on_project_id_ral_20", :algorithm=>:concurrently})
   -> 0.0026s
-- execute("SET statement_timeout TO 0")
   -> 0.0005s
-- add_index(:project_features, :project_id, {:where=>"repository_access_level = 20", :name=>"index_project_features_on_project_id_ral_20", :algorithm=>:concurrently})
   -> 0.0037s
-- execute("RESET ALL")
   -> 0.0004s
== 20200309140540 AddIndexOnProjectIdAndRepositoryAccessLevelToProjectFeatures: migrated (0.0074s)

✔ ~/projects/gdk/gitlab [208887-optimize-project-counters-projects_with_repositories_enabled ↑·129|✚ 1…1⚑ 1]
10:11 $ VERBOSE=true be rake db:migrate:down VERSION=20200309140540
== 20200309140540 AddIndexOnProjectIdAndRepositoryAccessLevelToProjectFeatures: reverting
-- transaction_open?()
   -> 0.0000s
-- indexes(:project_features)
   -> 0.0026s
-- execute("SET statement_timeout TO 0")
   -> 0.0005s
-- remove_index(:project_features, {:algorithm=>:concurrently, :name=>"index_project_features_on_project_id_ral_20"})
   -> 0.0039s
-- execute("RESET ALL")
   -> 0.0004s
== 20200309140540 AddIndexOnProjectIdAndRepositoryAccessLevelToProjectFeatures: reverted (0.0075s)

Does this MR meet the acceptance criteria?

Conformity

• Changelog entry
• Documentation (if required)
• Code review guidelines
• Merge request performance guidelines
• Style guides
• Database guides
• Separation of EE specific content

Availability and Testing

• Review and add/update tests for this feature/bug. Consider all test levels. See the Test Planning Process.
• Tested in all supported browsers
• Informed Infrastructure department of a default or new setting change, if applicable per definition of done

Security

If this MR contains changes to processing or storing of credentials or tokens, authorization and authentication methods and other items described in the security review guidelines:

• Label as security and @ mention @gitlab-com/gl-security/appsec
• The MR includes necessary changes to maintain consistency between UI, API, email, or other methods
• Security reports checked/validated by a reviewer from the AppSec team

PartOf #208887 (closed)