[Live-devel] Live555 scalability issue.

[Nuno Mota]

Thanks for the input I'll try to look into those issues.

Regards,
Nuno
2011/6/7 Ross Finlayson <finlayson at live555.com>

> a couple of days ago i conducted some scalability tests with the live555
>> media server. The computer in test was a quad core that handle the job
>> pretty well and of course a Gigabit port was used. For the last test I used
>> several openRTSP clients. The video source I used was a transport steam
>> file. The normal client downloads an average of 10Mbit/s (1,3Mbytes/s). For
>> the test I created 68 clients all connecting to the same live555 server. The
>> connection sequence was the following, 25 clients, 24 clients, 12 clients
>> and 8. For the first 49 clients the connection was stable and normal, having
>> a total throughput of more than 500Mbit/s. However when connecting the next
>> 12 and 8, the bandwidth was reduced more than 20% for every client i think.
>> Because testing for one  the average speed was less than 8Mbit/s with a
>> total Throughput from the server of around 500Mbit/s.
>>
>> Later when the first 25 clients disconnected and the rest remained, the
>> connection went up  reaching almost 750Mbit for 5 minutes and just for 44
>> clients, meaning they were compensating I guess for the stream delay. In the
>> end all the remaining clients had a video file almost identical to the
>> original one (missing a few seconds maybe). Why did this happen?
>>
>
> I don't know.  You'll have to explicitly measure exactly where in your
> system the bottleneck(s) are occurring, and exactly what is getting
> overloaded.
>
> Note that there's a lot more in 'your system' than just our software. In
> particular, you have operating systems, CPUs, networks, routers, etc.  Any
> of which could be limiting scalability.
>
> Note in particular that scalability problems with this (and other) software
> is often caused by operating-system-imposed limits on the number of open
> sockets.  Such a limit can usually be increased by reconfiguring your OS, so
> that's one of many things that you might try.
>
>
>
>  I'm guessing ofcourse this is due to the RTCP protocol.
>>
>
> No, that's highly unlikely.  The overhead of RTCP is negligible.
> --
>
> Ross Finlayson
> Live Networks, Inc.
> http://www.live555.com/
> _______________________________________________
> live-devel mailing list
> live-devel at lists.live555.com
> http://lists.live555.com/mailman/listinfo/live-devel
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.live555.com/pipermail/live-devel/attachments/20110607/5ff81805/attachment.html>