Open prologic opened 7 years ago
Hmm, something along the lines of maybe... connect 1000 clients, join them all to a single channel, and measure the latency of both sending the original message and having other clients receive it?
I believe so; You could even just do something like:
connect 1000 clients pair them up measure their CTCP TIME request/response(es)
Print out p59, p95 and p99 latencies.
James Mills / prologic
E: prologic@shortcircuit.net.au W: prologic.shortcircuit.net.au
On Wed, Nov 15, 2017 at 6:48 PM, Daniel Oaks notifications@github.com wrote:
Hmm, something along the lines of maybe... connect 1000 clients, join them all to a single channel, and measure the latency of both sending the original message and having other clients receive it?
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/DanielOaks/irc-stress-test/issues/1#issuecomment-344799041, or mute the thread https://github.com/notifications/unsubscribe-auth/ABOv-u-FG642pLYbnibhtan_NNdld6_gks5s26KVgaJpZM4Qf2qL .
I'd like this tool to be expanded to measure latency with many simulated concurrent clients connected. This will hopefully help measure (if you do some profiling on the ircd side) where all the time is being spent in constructing responses.