Closed essen closed 12 years ago
Sure thing. Tonight I'm adding a benchmark for EventMachine and Twisted's websocket implementation. I'll rerun the Cowboy benchmark using version 0.6.0.
You may want to look at the data for the Cowboy benchmark. 25% of the handshakes were more than 742ms. While the TCP connection didn't timeout, up to 2,500 of the connections may have had really slow handshakes. 5% of the connections had > 4sec handshakes.
Thanks what I get for tweeting at 3 in the morning :)
If it's with 0.6, I know where the problem lies, it's already fixed in my local code.-- Envoyé depuis mon HP Pre3Le 13 juin 2012 19:27, ericmoritz reply@reply.github.com a écrit : You may want to look at the data for the Cowboy benchmark. 25% of the handshakes were more than 742ms. While the TCP connection didn't timeout, up to 2,500 of the connections may have had really slow handshakes. 5% of the connections had > 4sec handshakes.
Reply to this email directly or view it on GitHub: https://github.com/ericmoritz/wsdemo/pull/1#issuecomment-6307556
This was based on 'master' as of last night around 1 or 2 am EST. I destroyed the AWS instances so I can't say for sure. I'll let you know what it looks like on 0.6.0 tonight.
Reply to this email directly or view it on GitHub: https://github.com/ericmoritz/wsdemo/pull/1#issuecomment-6311011
So, yes it was done on 0.6 then. ;)
I have breaking changes planned after 0.6.0 and I wouldn't want this benchmark to stop working considering how important it is. :)
Also if that's OK with you I will include a few slides about it in my OSCON talk in July.