This documentation describes some levels of production readiness. Please refer Production Readiness Checklist for checks which are required to pass to meet a level. Choose a level for a microservice based on SLO.
The main factor of choosing a Service Level is the expected SLO. i.e., You need to define your service SLO in order to choose your service level.
New to SLO? See guidelines from Google
Here's the typical required SLO for each level:
| Service Levels | Required SLO |
|---|---|
| A | > 99.9% |
| B | 99% ~ 99.9% |
| C | 95% ~ 99% |
When checking production readiness, you will need to go through the Production Readiness Checklist which requires different check items for different levels.
If you find yourself having a hard time choosing the right level, a general advice is to start from Level B and work on improving it to meet Level A if necessary.
Level A is for critical microservices. Services that handle PII, contribute directly and significantly to company-level goals (e.g., GMV), or could lead to significant damage to the company during downtime (e.g., anti-social check), should usually be Level A.
Level B is for standard microservices. This applies to services that provide necessary functionality, but won't lead to significant usability downgrade during downtime, or can be remedied without largely affecting company-level goals. Examples: like-service, notification-service.
Level C is for experimental microservices: services that provide experimental new features, or services that affect only a small part of users.