Serverless Benchmarks

[elthrasher]

A suite of tests that benchmark the same functionality with different architecture or technology choices (i.e. HTTP API vs REST API, Java vs Python, different SDK versions).

The goal of this is to assist developers to make technology choices.

• Track drift over time
• Easy to onboard new tests
• Roll up into aggregate results
• Report results

Further Context

• https://awscommunitybuilders.slack.com/archives/C031J0741UK/p1645893777189899
• https://github.com/elthrasher/serverless-benchmarks

Volunteers

@elthrasher
@JonHolman

[elthrasher]

I was rethinking this project a little and came up with a new architecture to support multiple Lambda functions and API Gateway invocations.

When it comes to metrics collections, it also makes sense to keep a running total maybe with indicators going up and down.

Should x-ray instrumentation be optional? Leaving that off for now, but it might be nice to be able to see if x-ray impacts performance and how much.

[JonHolman]

@elthrasher like I was saying in our chat. The data for whatever exercise we're going to have the lambda function act on in its test will need to be delivered as a part of the payload, so the API Gateway service proxy to DynamoDB use case can be tested.