Fault Tolerance - Serverless Architectures with AWS Lambda

Fault Tolerance

If the application availability you need requires you to take advantage of multiple AWS Regions, you must take this into account up front in your design. It’s not a complex exercise to replicate your Lambda function code packages to multiple AWS Regions. What can be complex, like most multi-region application designs, is coordinating a failover decision across all tiers of your application stack. This means you need to understand and orchestrate the shift to another AWS Region—not just for your Lambda functions but also for your event sources (and dependencies further upstream of your event sources) and persistence layers. In the end, a multi-region architecture is very application-specific. The most important thing to do to make a multi-region design feasible is to account for it in your design up front.