[bigbluebutton-users] Results From BigBlueButton Stress Test

[Jeremy Thomerson]

First of all, thank you to everyone who joined us for the stress test earlier this week.  Your efforts will assist us in making BigBlueButton a better product.

Attached to this email, you will find a PDF containing a couple of charts: CPU and network utilization during the test.  You will likely notice that the CPU is half of what you may expect.  This was a dual core machine - and it appears that Red5 is only utilizing a single core.  That's a valuable piece of information - and something we can look at improving.  

We used only the default BigBlueButton audio during the conference.  This really seemed to be the only pain point.  When you use direct phone access to Asterisk (which is what we do at Generic Conferencing), you get high-quality audio.  But the audio through the Flash -> Red5 -> Asterisk -> Red5 -> Flash route degrades the quality noticeably, especially when you start adding more users in.  Of course, we already knew this.  And Fred, Richard, and the rest of the Blindside Networks team are already working on several things that will improve this over time.

The webcam performance really didn't seem to degrade for us until we had well over 130 webcams streaming.  And, this will most likely even improve in the future if we are able to transcode the incoming video on the server before broadcasting it.

I guess my takeaway would be that BigBlueButton can easily handle 50 users.  And, with most conferences, where you would only have one to four people talking and sharing webcams, BigBlueButton could probably support broadcasting this audio and video to 100 users.  

Hopefully this is helpful to everyone.  We'll be planning on scheduling another stress test for upcoming releases.  We hope we'll be able to have even more unique users for those.

Jeremy Thomerson

http://genericconf.com

BigBlueButton with local numbers and high-quality audio - conferencing that's better than free!

--
You received this message because you are subscribed to the Google Groups "bigbluebutton-users" group.
To post to this group, send email to bigbluebu...@googlegroups.com.
To unsubscribe from this group, send email to bigbluebutton-u...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/bigbluebutton-users?hl=en.

[Jeremy Thomerson]

I forgot my screenshots!  Here are two - and yes that last one says 163 users!  We had about thirty unique users who each had a BUNCH of windows open and webcams all over the place!

Jeremy

On Fri, May 7, 2010 at 3:04 PM, Fred Dixon <ffd...@gmail.com> wrote:

Thanks Jeremy for sharing these results!

I took a few screen shots during the session for those who couldn't
attend.   The last screen shot show the "let's go for it" where the 20
users each opened multiple instances of BigBlueButton, peaking at 172
simultaneous users.

Regards,... Fred

--
You received this message because you are subscribed to the Google Groups "BigBlueButton-dev" group.
To post to this group, send email to bigblueb...@googlegroups.com.
To unsubscribe from this group, send email to bigbluebutton-...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/bigbluebutton-dev?hl=en.

[Guillermo Vargas-Dellacasa]

Thank you very much for sharing this. It is truly interesting
information for those who follow BBB but couldn't participate. May I
ask a couple of questions:

What were the characteristics of the server hosting the conference:
processor type, speed, memory, and OS (I assume Ubuntu 32bit)?

Also, in your opinion, would the load seen on the server have been the
same had the test consisted of 8 simultaneous conferences of 20
participants each (8x20) instead of a single conference of 160 users
(1x160)?

Again, thank you very much for sharing this with the community.

Guillermo

On May 7, 4:10 pm, Jeremy Thomerson <jeremythomer...@gmail.com> wrote:
> I forgot my screenshots!  Here are two - and yes that last one says 163
> users!  We had about thirty unique users who each had a BUNCH of windows
> open and webcams all over the place!
>
> Jeremy
>
>
>
>
>
> On Fri, May 7, 2010 at 3:04 PM, Fred Dixon <ffdi...@gmail.com> wrote:
> > Thanks Jeremy for sharing these results!
>
> > I took a few screen shots during the session for those who couldn't
> > attend.   The last screen shot show the "let's go for it" where the 20
> > users each opened multiple instances of BigBlueButton, peaking at 172
> > simultaneous users.
>
> > Regards,... Fred
>
> > --
> > You received this message because you are subscribed to the Google Groups
> > "BigBlueButton-dev" group.
> > To post to this group, send email to bigblueb...@googlegroups.com.
> > To unsubscribe from this group, send email to

> > bigbluebutton-...@googlegroups.com<bigbluebutton-dev%2Bunsubscr i...@googlegroups.com>
> > .

> > For more options, visit this group at
> >http://groups.google.com/group/bigbluebutton-dev?hl=en.
>
> --
> You received this message because you are subscribed to the Google Groups "bigbluebutton-users" group.
> To post to this group, send email to bigbluebu...@googlegroups.com.
> To unsubscribe from this group, send email to bigbluebutton-u...@googlegroups.com.
> For more options, visit this group athttp://groups.google.com/group/bigbluebutton-users?hl=en.
>

>  bbb-stresstest-4.png
> 1399KViewDownload
>
>  bbb-stresstest-6.png
> 1479KViewDownload

[Jeremy Thomerson]

It is a desktop-class machine.  It has the following specs:

Pentium(R) Dual-Core  CPU      E5200  @ 2.50GHz

4GB RAM

Ubuntu 9.04 32-bit

Yes, I think the load would be very equivalent spread across multiple conferences.  Perhaps slightly different, but fairly equal.

Jeremy