Test contour and public cloud ingress comparison

[iyacontrol]

how to optimize contour？ i use ab to test contour and elb.

result: contour:

Server Software:	envoy
Server Port	80
Document Path	/
Document Length	612 bytes
Concurrency Level	10
Time taken for tests	12.048 seconds
Complete requests	100
Failed requests	0
Total transferred	87200 bytes
HTML transferred	61200 bytes
Requests per second	8.30
Transfer rate	7.07 kb/s received

cloud ingress :

Server Software:	elb
Server Port	80
Document Path	/
Document Length	612 bytes
Concurrency Level	10
Time taken for tests	1.828 seconds
Complete requests	100
Failed requests	0
Total transferred	83600 bytes
HTML transferred	61200 bytes
Requests per second	54.70
Transfer rate	44.66 kb/s received

Environment:

• Contour version: 1.2.1
• Kubernetes version: 1.13:

[jpeach]

I spent some time replicating this.

I spun up a 4 node cluster on GKE with this script. My backend service was this echo server, which ab can drive at 32K RPS locally on my laptop. I did a variety of simple tests. When I run ab, I am doing ab -k -c 4 -t -n 9999999.

RPS	Description
9000	Running both ab and echo server in the same container. Networking completely over loopback.
3000	Running ab inside the envoy container, so the first hop is loopback, the second hop is the cluster network.
200	Running ab from my laptop to the echo server running behind a GKE load balancer (TCP balancer).
170	Running ab from my laptop to the echo server running behind envoy. This goes to the GKE load balancer, to envoy, then across the cluster network.
4	If you force envoy to generate a 404 it seems incredibly slow.

Note that this is all pretty quick & dirty and is only interesting for the comparison.

[github-actions[bot]]

The Contour project currently lacks enough contributors to adequately respond to all Issues.

This bot triages Issues according to the following rules:

• After 60d of inactivity, lifecycle/stale is applied
• After 30d of inactivity since lifecycle/stale was applied, the Issue is closed

You can:

• Mark this Issue as fresh by commenting
• Close this Issue
• Offer to help out with triage

Please send feedback to the #contour channel in the Kubernetes Slack

[github-actions[bot]]

The Contour project currently lacks enough contributors to adequately respond to all Issues.

This bot triages Issues according to the following rules:

• After 60d of inactivity, lifecycle/stale is applied
• After 30d of inactivity since lifecycle/stale was applied, the Issue is closed

You can:

• Mark this Issue as fresh by commenting
• Close this Issue
• Offer to help out with triage

Please send feedback to the #contour channel in the Kubernetes Slack