Skip to content

Conversation

DeagleGross
Copy link
Contributor

changed to the same client and same request sending to both kestrel & http.sys and both are responding in same manner. shows kestrel being a bit slower than http.sys (85k RPS vs 90k RPS) on perf lab machines.

Socket errors are reporting 2x for Kestrel due to client specifics. we dont care really: the only interesting value is RPS & that all requests are not successful

kestrel:

| Max Global CPU Usage (%)  | 81                  |
| Max Process CPU Usage (%) | 80                 |
| Max Global CPU Usage (%)  | 89                  |
| Connections               | 32                  |
| Bad responses             | 83,051              |
| Requests                  | 236,919             |
| Requests/sec              | 5,535               |
| Socket Errors             | 153,868             |

http.sys:

| Max Global CPU Usage (%)  | 84                 |
| Max Process CPU Usage (%) | 1                 |
| Connections               | 32                  |
| Bad responses             | 90,879              |
| Requests                  | 181,762             |
| Requests/sec              | 6,055               |
| Socket Errors             | 90,883              |

@DeagleGross DeagleGross self-assigned this May 23, 2025
@DeagleGross DeagleGross reopened this May 23, 2025
@DeagleGross DeagleGross merged commit 302fb36 into aspnet:main May 23, 2025
2 checks passed
@DeagleGross DeagleGross deleted the dmkorolev/rejection-header branch May 26, 2025 09:04
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants