Layer7 API Management

Hi There,

We are doing a performance testing with a load of 750 TPS with 2 cluster with nine 900.Gateway relaying on back-end for all this requests, if that was the case based on this how much read time out would it be recommended to set on the route in connections tab? we are setting right now 10,000 milliseconds which is of 2 sec and we were getting a read time out issue(-4) from back-end. How we tackle it and what is recommended read time out can be set ? pls advice,

Hello,

I believe everybody wants to see all requests are processed successfully without TIMEOUT, so the best way is to make the back-end respond quickly enough and have plenty of concurrency for satisfying the 750TPS.
If the back-end is unchangeable, I think the second best would be to limit the number of incoming requests to the API Gateway (750TPS won't be satisfied though).

Probably the read time out should be decided from the longest response time of the back-end (the latency of the back-end API).
If the read time out is shorter than the longest response time, the API Gateway gives up to receive the response from the back-end even though it is being prepared.
If the read time out is too much longer than the longest response time, the API Gateway will be wasting time even if the back-end will never respond.

I hope this answers your question.

Best regards,
Seiji