Operational excellency

The DevOps team should be given the mandate to automate as much of the operational tasks as possible.

Uptime monitoring for API consumers

In the API portal, have a common pattern for API uptime monitoring where the focus group is API consumers. For example, each API could have a health endpoint which performs an end-to-end call and has a standard payload.

Have an uptime status page where all the health statuses are presented. Overtime, this can be extended to provide alerts based on a publish/subscribe pattern.

Uptime monitoring for API producer

Do automated tests at least every hour through a background script and send alerts on failure to the DevOps team (for example through slack) It should be a clear goal that the DevOps team should be aware of issues and working on resolving them, before it is reported by API consumers.

Operational issues

Prepare a process for how to handle operational issues which require an emergency change. Include responsibilities and stepwise explanation on how to continously inform affected API consumers of the progress.

Monitoring of users with missing authorization

A very common error in internal APIs, is that the end-user is missing some rights in the back-end system. Provide detailed error messages in these cases and work with API consumers on how to reduce the likelihood of such cases.

Expiring keys and certificates

Have a calendar for expiring keys and certificates and make sure their are update well in advance of the expiry.