I'm deploying a serverless/lambda-based HTTP API via AWS API Gateway. My goal is to minimize latency in handling requests. I am considering two architectures, and I'm wondering which best meets this goal. Here are the options:
Deploy one edge-optimized API Gateway application. AWS will then create a CloudFront distribution which will route incoming requests through the AWS Edge Network of global locations, but ultimately requests will be handled at one specific deployment location (in the US).
Deploy multiple regional API Gateway application instances (one in the US, a second in Europe, and a third in Asia), and then use latency routing to route requests to the region that provides the best latency with less round-trip time.