Skip to content

Assignment 4 Testing Strategy

Krishna Vamsi Guntupalli edited this page May 7, 2022 · 3 revisions

It took long time for us to complete the deployment of custos. So we could not properly test it.

Goal: We tried to load test to check when custos setup breaks with different number of users and ramp-up times.

Results:

WhatsApp Image 2022-05-07 at 12 22 56 AM

WhatsApp Image 2022-05-07 at 12 22 52 AM (1)

WhatsApp Image 2022-05-07 at 12 22 52 AM (2)

WhatsApp Image 2022-05-07 at 12 22 52 AM (3)

WhatsApp Image 2022-05-07 at 12 22 52 AM (4)

WhatsApp Image 2022-05-07 at 12 22 52 AM

WhatsApp Image 2022-05-07 at 12 22 53 AM

WhatsApp Image 2022-05-07 at 12 22 53 AM (1)

We did load testing with different number of users and ramp-up times to check if custos setup would break. We tested up to 16 endpoints with 100 concurrent users and 1 seconds ramp up time. While testing we made the following observations. We observed the user service pods are having more failure rate. May be increasing the memory limit of this services would resolve this problem. Most micro-services in custos architecture depend on a single microservice - custos config service which can act as a single point of failure. We can scale up the custos config services or divide the functionalities into multiple microservices so that custos config service does not act as a single point of failure.

Being new using this application, it took more time than we predicted to deploy custos on jetstream. Owing to the delay in deployment, we did not have enough time to implement stress testing and other strategies which we learnt as part of this course. Only drawback which we felt is, if the documentation has been provided with more examples it would have been helpful for users like us who are relatively new to it.

We would like to again thank Team Terra who have been very helpful in completing this assignment successfully.