Bachmarks

How to run benchmark

go test -bench .

Report

The overhead of the policy enforcement is benchmarked by eval_b_test.go. And the report was generated by running the benchmark cases on the testbed:

2 * Intel(R) Xeon(R) Platinum 8167M CPU @ 2.00GHz 2 Cores, 4 Threads

This tables shows the report for API IsAllowed. op means one call of IsAllowed.

Case	Size	Time overhead (μs/op)
One	1 policy, 1 role policy, 1 user	6.6
Tiny	10 policies, 10 role policies, 10 users/role	7.3
Small	100 policies, 100 role policies, 10 users/role	8.5
Medium	1K policies, 1K role policies, 10 users/role	8.6
Large	10K policies, 10K role policies, 10 users/role	8.7
Huge	100K policies, 100K role policies, 10 users/role	8.7
LargeExp	10K policies with resource expression, 10K role policies, 10 users/role	43.0
LargeCond	10K policies with condition, 10K role policies, 10 users/role	8.0

And below table shows the report for API GetAllGrantedPermissions. op means one call of GetAllGrantedPermissions.

Case	Size	Time overhead (μs/op)
LargePerm	10K policies, 10K role policies, 10 users/role	8.9