Define and set up a 100k reference architecture configuration

Example customer requirements:

• Zero Downtime upgrades
• Scale - 100k users, 300K repos
• Low latency - significant distribution globally
• 2000 RPS

Acceptance criteria for certification (tentative):

• Latencies at 2000 RPS on the 100k environment should be less than 500ms, with the exception of known bugs affecting certain endpoints.
• Endpoints that have known bugs have issues created and fed into the Performance Grooming process.
• Documentation has been added to describe the 100k environment's configuration.

Diagram

• Google Drawing TBD

Documentation

• TBD https://docs.gitlab.com/ee/administration/high_availability/