0015 - Multi Region by Default

Date
Authors
John Nolan
Status
accepted
Areas

Context

We want our services to be resilient to data center failure. If a data center or region goes down, we want to be able to switch to a new region quickly with minimal to no data loss.

Region failure is rare but does happen. Common situations for region failure are

  • Natural disaster
  • DDOS attacks
  • Data Center outages

Decision

Our other services already are or are in the process of using Multi Region Architecture.

We should continue to use this strategy for any new architecture.

Consequences

Positive

  • It is quicker to build our services as multi region from the start rather than retroactively
  • We can fail over to another region in case of an outage
  • If we are unable to use a particular region due to Government policy changes, we can quickly move

Negative

  • Initial building times are slightly longer
  • Developers will need to factor in multi region when designing their services