wrk pressure test，8 cpu 16g ram，qps below 100k?

[laughingyear]

I chang apisix/conf/config.yaml ,add below to use port 7890:

nginx_config:
main_configuration_snippet: |
daemon on;
http_configuration_snippet: |
server
{
listen 7890;
server_name _;
access_log off;

        location / {
            default_type text/html;
            return 200 'This is text!';
        }
    }

    chunked_transfer_encoding on;

i add apisix route:

curl "http://127.0.0.1:9080/apisix/admin/upstreams/1" -H "X-API-KEY: edd1c9f034335f136f87ad84b625c8f1" -X PUT -d '
{
"type": "roundrobin",
"nodes": {
"127.0.0.1:7890": 1
}
}'

curl "http://127.0.0.1:9080/apisix/admin/routes/1" -H "X-API-KEY: edd1c9f034335f136f87ad84b625c8f1" -X PUT -d '
{
"uri": "/",
"host": "127.0.0.1",
"upstream_id": "1"
}'

then i use wrk:
wrk -t8 -c10000 -d30 --latency http://127.0.0.1:9080/
Running 30s test @ http://127.0.0.1:9080/
8 threads and 10000 connections
Thread Stats Avg Stdev Max +/- Stdev
Latency 39.25ms 161.09ms 1.97s 96.44%
Req/Sec 10.24k 4.24k 25.63k 65.88%
Latency Distribution
50% 9.36ms
75% 16.10ms
90% 29.94ms
99% 1.03s
2446920 requests in 30.10s, 410.63MB read
Socket errors: connect 0, read 0, write 0, timeout 10656
Non-2xx or 3xx responses: 157
Requests/sec: 81296.78
Transfer/sec: 13.64MB

command top shows:

[tzssangglass]

Is your upstream, APISIX and wrk all on the same machine? Perhaps wrk and upstream are competing for system resources.

[laughingyear]

Is your upstream, APISIX and wrk all on the same machine? Perhaps wrk and upstream are competing for system resources.

yeah，but cpu is only 60%

[tzssangglass]

What is your expected performance?

[laughingyear]

What is your expected performance?
I think it should be 140K qps

[tzssangglass]

I would like to learn how this value is calculated.

BTW, here is the benchmark way: https://github.com/apache/apisix/blob/master/docs/en/latest/benchmark.md