Closed hectorj2f closed 7 years ago
I started to get some results in this document https://docs.google.com/a/redhat.com/document/d/1gcuQbEL2ZTiVcmy39EFMVV9fZxCaS3Oj6ikKZHk9q18/edit?usp=sharing and in this repository https://github.com/hectorj2f/keycloak-perf-tests
TODO: I'll extract an average of these executions and conclude with few graphs to easily explain the performance spectrum of KC in our scenario.
@alexeykazakov Do you have better ideas or anything else we might want to test for KC ? Otherwise I'll close this issue.
I'll run two more tests one against a RDS database instead of a postgresql and one test against our SSO cluster.
What you can also think of is to tune JDBC connection pooling etc. Also @mposolda might have some insights. Would make sense to reach out to him at some point.
I updated the document with some new experiments:
I share the link to a new document about the performance analysis https://docs.google.com/a/redhat.com/document/d/1nXzEE_lPf6tmsJLcVu5l7O59KcOS4rNUYb8oB6KIQvI/edit?usp=sharing
We ran several performance tests and extracted some conclusions. The intention was to reduce the traffic from core to those endpoints that suffer more with a high workload. There is another project fabric8-auth which would try to provide other functionalities to avoid using KC for those, thereby reducing the workload as well.
We want to evaluate the performance behavior of a Keycloak cluster.