BigBlueButton (web conferencing system) -> What root server will suffice?

  • Good day.


    I was wondering if you could help me out.


    I"m looking for a server to run BBB.


    Here is the thing:


    1) The devs mention that VPS is not recommended (see the link below). I read their explanation and followed the link they provided, but tbh I'm not convinced.


    2) The installpage contains specific requirements ( 4 core etc), however, on BBB forums I read that

    Quote

    I have run BBB on Intel Pentium 2.4ghz with 2gb ram and did 20concurrent(with few video streams) successfully"

    Our distance learning student base is pretty low, for the time being, about it's about 20-30 users.


    I'd like to know whether RS 1000 would run BBB great despite having just 2 cores or should I get RS 2000?



    Thank you in advance


    PS And another question...


    Do coupons get applied to recurring payments or it's just for the first payment?


    The description says "The voucher adds a new vServer rate with a reduced monthly amount to your shopping cart "


    but I just wanted to be sure.


  • Hay,


    I don't know the software, but the part, which you warn about vps, is freeswitch - that is a telephony module, comparable to the asterisk pbx.


    With the asterisk, I can make 1000 parallel calls (maybe more) on my RS8000 server and I don't think there will be a problem with a RS2000 with less calls. Please be aware that the RSxxxx-Servers are also virtualized, but with dedicated ressources (RAM, CPU). The mentioned "bare metal server" is a really dedicated server under https://www.netcup.de/professional/.


    (It's funny, that freeswitch don't recommend virtual servers, but provides a full installed virtual machine for virtualbox. And recommends the AWS-platform too, which is also virtualized.)


    In other parts, I would follow the recommendations, it seems to be a very special software.


    CU, Peter

  • that freeswitch don't recommend virtual servers

    Never seen in the docs that FS is recommending metal.

    The "problem" lies in the re-coding of voice and video - if codecs on both ends don't match - then freeswitch acts as a media server to translate between codecs.


    On my small Fujitsu Futro systems I get 20 concurrent calls without any problems. On my FS development virtual machines I don't see any problems at all. So the RS 1000 should be enough to run 20 concurrent calls.


    CmdrXay with FreeSWITCH Blue they provide fully configured FS virtual machines. https://freeswitch.com/index.php/fs-blue/

  • Do coupons get applied to recurring payments or it's just for the first payment?

    If you have a 5€ coupon that one is applied only once.

    If you have a 10% coupon that one is applied monthly.


    If the RS 1000 doesn't provide enough power for your application, you may give the server back for free and take a RS 2000.

    But I don't know whether that offer is only for customers from Germany or for international customers too.

  • Hay,


    H6G  https://freeswitch.org/conflue…RecommendedConfigurations

    Quote

    A 64-bit CPU running a 64-bit operating system and a 64-bit version of FreeSWITCH is recommended. A bare metal system provides consistent, predictable performance and most importantly for real–time applications like this, a reliable kernel clock for RTP packet timing. With a virtual machine it is difficult to determine where any problems might originate and improper propagation of the hardware clock through the VM host to the guest operating system is not always available so the RTP tests will be rendered meaningless.


    That can be read as "please dont use virtual systems". I can't follow this from my own experience.


    In my first asterisk-system (its not really comparable, but its the same with encoding and voice streams) I had 2 x 1.4 GHZ Pentium IIIS CPUs running and 100 parallel calls without latencies. On the AWS I had 10 of these c4.4xlarge (16 vCPU/30 GB RAM) and we had 50.000 concurrent calls (with reserve). We tested the t3.micro (2 vCPU/1 GB RAM) with 400 (!!) concurrent calls without problems.


    So, Im sure that RS1000 or RS2000 will fit your use case.


    CU, Peter

  • Hey guys,


    I know this in an old thread but I just wanted to share my experience with running BBB (freeswitch) on RS.(6 cores dedicated)


    I had audio distortions and I couldn't find what caused them. I tried changing specific freeswich settings in switch.conf.xml, conference.conf.xml etc but nothing helped.

    I suspect it had something to do with timing issue coz of virtualization environment.


    I had them on any VPS I tried running BBB on. It was considerably better on netcups's RS servers though compared to ordinary VPS with shared cores.


    After I have switched to a bare metal server these distortions disappeared completely.


    If anyone here had experience with running freeswitch on RS I would appreciate if you shared it here coz honestly I would prefer using netcup's RS instead of a bare metal server. I just love the ease of making snapshots too much lol.


    Thank you.

  • The CPU cores of all RS are dedicated. The difference to VPS is, that their virtual cores are shared.

    In your test case (bare metal server) you probably eliminated network latency and possible packet loss or even MTU problems over your leased line, too. So, I guess, your question isn't leading you to the right direction in the first place.


    Do you have more information you'd like to share with us?

  • Yeah I know the difference between general vps and vps with dedicated cores.=)

    you probably eliminated network latency and possible packet los

    Nah, I didn't have any latency issues or packet loss with netcup's RS4000 server.



    MTU problems

    Hmm I don't think so.


    Do you have more information you'd like to share with us?

    Tbh I was hoping someone here had some experience with running freeswitch on a virtual server without audio distortions.

  • Tbh I was hoping someone here had some experience with running freeswitch on a virtual server without audio distortions.

    Usually, distortions are caused by network problems. I wouldn't suspect the CPU in the first place. You could do some monitoring, collecting statistics of CPU usage, I/O and packet loss. Some statistics are already available on your "servercontrolpanel". Look, if you can see excursions of the graphs, that would correlate to a distortion event, you've observed.


    I had them on any VPS I tried running BBB on. It was considerably better on netcups's RS servers though compared to ordinary VPS with shared cores.


    After I have switched to a bare metal server these distortions disappeared completely.



    Nah, I didn't have any latency issues or packet loss with netcup's RS4000 server.


    I don't get the point. You didn't have latency issues or packet loss on the RS and you are still experiencing distortions? What kind of distortions exactly? dropouts, jitter?


    Remember, that VoIP telephony usually used UDP or RTP, so transmissions are not secured against packet loss, which is most likely to appear on your ISPs leased line. When you write, the problem does not persist on a bare metal server, I suppose that server is in your LAN. Isn't it?

    In that case, you'd be mixing up reason and effect.

  • Quote

    You didn't have latency issues or packet loss on the RS and you are still experiencing distortions?


    Yes.

    Quote

    What kind of distortions exactly?

    Audio gets slowed down and distorted. it's not constant but it happens often enough to be a serious issue.


    Quote

    used UDP or RTP, so transmissions are not secured against packet loss


    I already said there was no packet loss, no jitter. I tested on different networks(client side) and different vps servers,


    Quote

    I suppose that server is in your LAN. Isn't it?


    No. It' s a bare metal server in a data center



    If you google freeswitch + virtual server\vps you'll see that users are likely to experience issues with vps and freeswitch. They happen often enough for the freeswitch devs to specifically recommend using bare metal servers.


    Some manage to make it run OK though.


    I thought the problem was with the freeswitch internal timer or something but changing corresponding settings didn't help.


    PS CPU load is less than 5-10%, no I\O issues or network issues. It's interesting to note that switching from a pure VPS to a VPS with dedicated cores significantly reduces these distortions. So, the problem is related to virtualization. I just wish i knew if it can be fixed via changing some settings


    Regards