Infrastructure planning starts with the application and the people who depend on it. Record the data it stores, the services it calls and the consequences of an interruption.
Define the operating requirements
List the application, database, file storage and scheduled jobs. Describe expected traffic and identify busy periods. Agree on acceptable downtime and data loss so the hosting plan has clear requirements.
Make releases repeatable
Document how a tested change reaches production and how a previous version can be restored. Keep credentials out of the public website folder. Check database changes as part of the release plan, including what happens if a deployment is interrupted.
Verify recovery
AWS guidance recommends backing up the data needed to meet recovery requirements and testing restoration. Include uploaded files and configuration needed to rebuild the application, alongside the database. Record who can restore a backup and practise doing so in a separate environment.
Give alerts an owner
Decide which failures require action and who receives the alert. Review application errors, failed jobs, storage capacity and resource use. An alert should point to a documented response.
- Keep a list of critical services and contacts.
- Review backup and restore results.
- Review usage and operating costs regularly.
Further reading: AWS: Back up data