Hi, We are facing an issue of high response time when test is running with as low as 5 RPM, but the application is consistent with 2K, 10K RPM.

Google Cloud Apigee
1

Architecture : Api Gateway (AWS us-west) → Service Layer (AWS us-east) → 3rd party API (AWS us-east)

During low RPM test,

Nw Latency between Api Gateway - Service Layer is as high as 280 msec

Nw Latency between Service Layer - 3rd party API is as high as 150 msec

Api Gateway → 40 msec

Service Layer → 5 msec

3rd party API → 300 msec

Due to this Total Response time is around 770 msec

During High RPM test,

Nw Latency between Api Gateway - Service Layer is as high as 80 msec

Nw Latency between Service Layer - 3rd party API is as high as 50 msec

Api Gateway → 10 msec

Service Layer → 5 msec

3rd party API → 140 msec

Due to this Total Response time is around 285 msec

Service is running on EKS cluster on AWS. Any pointers to debug on the n/w latency is much appreciated.

2

A test of 5 RPM, as in a single request every 12 seconds? I’m assuming you have a very small sample data set for this test, and it’s not clear what the other aggregations look like since you’ve mentioned “as high as x msec”. The smaller the sample size, the less consistent it’s going to be. I don’t think it’s a very useful test

Additionally, you should look at average times, and for example 90%/95% response times rather than what seems to be focussing on outliers on their own.

1 Like