Michael Penick
Michael Penick
Thanks for all the info. I'll try to reproduce and see what I can find. > The size can vary from extremely small to **megabyte** streams This is something I...
I ran some tests using 1MB payloads (reads) @ ~70 request/s. Here are the results (%iles in microseconds): #### direct (no proxy): ``` num_requests, duration, final rate, min, mean, median,...
I'm currently running a 1hr test @ ~70 request/s using cql-proxy. I'll post the result when done. Results: ``` num_requests, duration, final rate, min, mean, median, 75th, 95th, 98th, 99th,...
I noticed this: https://github.com/datastax/cql-proxy/commit/ea8421bab88bb6b85c6c22a1e1f010f5f955b026, Seeing a huge difference in performance (from reduced system calls). Give that a try, it's now on `main`. I'll also be pushing a release tomorrow.
> @mpenick Do you have any update/suggestions? I haven't been able to reproduce the issue yet so it's hard for me to know how to proceed here. I've tried to...
Thinking about this a bit more, maybe it makes sense to add some initial metrics to cql-proxy. We could then see where the extra latency is coming from by pointing...
Sorry for the delay. Thanks for the PR. Looks great, very thorough. Calling `SSL_CTX_set_default_verify_paths()` as default internally would be a bad idea because it might add CAs the application does...
Thanks for the test and trying to make it work on all platforms. I'll take a look.
I worked on this a bit yesterday on my local Windows machine. I thought I found the issue, and I'm able to get it working locally, but when I push...
Note: It works locally with many different versions of VS.