-
Bug
-
Resolution: Unresolved
-
Major
-
None
-
4.18.0, 4.19.0
-
Quality / Stability / Reliability
-
False
-
-
None
-
None
-
None
-
None
-
None
-
None
-
None
-
None
-
None
-
None
-
None
-
None
-
None
-
None
Description of problem:
We recently compared the ingress-performance of Azure with other cloud platform(AWS) and noticed the throughput in terms of request per sec is very low than AWS running similar cluster version.
Recent jobs from prow - perf dash (viewer/viewer)
Version-Release number of selected component (if applicable):
Noticed this on 4.18 and 4.19, could be on older releases as well
How reproducible:
Always
Steps to Reproduce:
1. Create a self-managed Azure cluster run ingress-perf workload - https://github.com/cloud-bulldozer/e2e-benchmarking/tree/master/workloads/network-perf-v2 2. Create AWS cluster of similar version and using eqiuvalent instance types run the same workload 3. Compare performance
Actual results:
Performance is worse by ~40%
Expected results:
Should be close with comparable deviation(10%) in numbers
Additional info:
Opened this bug to capture if there is any architectural difference between Azure OCP vs AWS OCP, if not lets try to improve it to perform equally.