Presence and IntComm WG

[Matthew Wild <>]

2009/11/5 Jorj Bauer 
<>:
> Notes from today's call. Our next scheduled call will be on Thursday,
> November 19th.
>
> Attendees: Tom Scavo, Jorj Bauer
>
> Tom notes that my announcement said we meet every Thursday, when we in fact
> meet the first and third Thursdays of the month. Noted, and I'll correct the
> announcement.
>
>> * [Jorj] Document Wave setup over OpenFire
>
> This documentation, as well as how to sit it on Prosody, should just be a
> pointer to Google's docs which are now evolving well. I'll dig up the
> relevant URLs and we can figure out where to put them in our wiki space.
>
>> * [Jorj] PICdemo dev system discussion
>
> I need to re-raise the development server question to Internet2, which I'm
> doing via Elaine right now. Once we have a development server we can stand
> up our own wave server there for testing purposes. Wave servers have no
> access control, so there is some risk involved; better to have that risk
> isolated to a development system.
>
>> * [Jorj] XMPP performance testing results
>
> Short synopsis of progress thus far: I'm testing ejabberd, prosody, and
> openfire.
>
>  - ejabberd is the best performer (expected)
>  - prosody is second
>  - openfire is third
>  - all three crash when you hit them too hard
>  - openfire's crashes are pathological; process doesn't exit, but hangs
>  - prosody's development is moving forward quickly and 0.7 may be worth
> recommending
>

Some notes from the Prosody perspective...

From the screencasts you made I note that Prosody stabilised at 16ms
latency, and ejabberd at 17ms latency, 1ms can make all the difference
;-)

I recall that the main issues you had with Prosody were that modifying
the contact list is not fast (on the plus side, reading is very fast,
which is the most common case). However if you really do need
thousands of contact list modifications each second then I guess the
upcoming support for different data backends will help a lot.

The second issue was with connect/disconnect taking some time with
lots of clients. I'm really hoping this will be much better with the
new connection framework landing in 0.7, so would be glad if I could
poke you to re-run your tests then.

I'm curious when you say that all three servers crashed under load...
exactly in what manner did Prosody "crash"? or do you simply refer to
the slowness in dealing with connections?

Overall your research was *very* helpful to us, so I'd like to say thanks!

Matthew