Ingress Performance Test Report¶
Test Background¶
Verify that the throughput of Ingress-nginx can increase linearly after horizontal scaling, and understand its performance metrics to help users manage resources effectively.
Benchmark¶
This article is based on the installation of Ingress-Nginx in DCE 5.0, with all parameters set to default configuration and without enabling features such as keepalive.
Physical Machine¶
Name | CPU | CPU | MEM | Interface |
---|---|---|---|---|
Node 1 | Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz | 56C | 128G | 10G Mellanox |
Node 2 | Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz | 56C | 128G | 10G Mellanox |
wrk Client | Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz | 56C | 128G | 10G Mellanox |
The following data is obtained from load testing on physical servers.
Number of Ingress-nginx CPUs | (wrk) Requests/sec |
---|---|
1 | 3209.66 |
2 | 5709.07 |
4 | 9005.79 |
8 | 20696.19 |
Relationship between Ingress-nginx CPU and RPS¶
Virtual Machine¶
Name | CPU | CPU | MEM |
---|---|---|---|
Node 1 | Intel(R) Xeon(R) Gold 5118 CPU @ 2.30GHz | 16C | 16G |
Node 2 | Intel(R) Xeon(R) Gold 5118 CPU @ 2.30GHz | 16C | 16G |
wrk Client | Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz | 56C | 128G |
The following data is obtained from load testing on virtual servers.
Number of Ingress-nginx CPUs | (wrk) Requests/sec |
---|---|
1 | 3022.68 |
2 | 5556.20 |
4 | 8616.95 |
8 | 11006.89 |