Can / AWA API Gateway & # 8594; improve lambda performance?

Question

Has anyone found a solution to Gateway API latency issues?

With a simple test of the Gateway → Lambda API, I regularly see cold starts in the range of 2.5 s, and as soon as it “warms up”, a response time in the range of 900ms - 1.1s is typical.

I understand that the TLS handshake has its own overhead, but testing similar resources (based on AWS or shared sites, which I believe are not geographically distributed) from my location shows results that are half less than ~ 500 ms.

Good news soon from AWS?

(I read everything I could find before posting.)

Answer 1

with the API Gateway team.

You said you read “everything,” but for context to others, I want to refer to several threads on our forums, where I publicly posted, where a lot of this perceived latency when making one API call comes from:

• Forum Post 1
• Forum Post 2

In general, as your call rates increase, the average latency will decrease as the mechanisms for reusing connections between your customers and CloudFront, as well as between CloudFront and the Gateway API, can be used. In addition, a higher call speed ensures that your Lambda is warm and ready to serve requests.

That being said, we are very confident that we do not meet the performance panel for many of our customers and are doing everything possible to improve this:

• The Lambda team is constantly working to improve cold start times, and is also trying to remove them for features that see continuous load.

• At Gateway API, we are currently in the process of deploying enhanced reuse of connections between CloudFront and Gateway APIs, where clients can benefit from connections established through other APIs. This should mean that the percentage of requests requiring full TLS messaging between CloudFront and the Gateway API should be reduced.