Results 1 to 5 of 5

Thread: Large cluster and high latency

  1. #1
    Join Date
    Aug 2011
    Posts
    6

    Default Large cluster and high latency

    Hi there,

    We have 14 streaming servers, each with 24GB ram, monster cpu's and 16X2TB raid-10 mounted HD.

    Each servers handles about 3.5k streams. Right now the problem we are facing is that the server are in north america and that we have listeners in europe who are experiencing issues like the video stop and auto-rewind 1 secon behind and then restart. Stream takes about 10 seconds to start and just stop to buffer etc ..

    We would really appreciate any hints on path to look at ..

    We dont think its only because of the cross-atlantic link.

    Our videos are encoded at a 375Kbps/second rate + the 96kbps audio. All of that done using ffmpeg.

    Thanks

  2. #2
    Join Date
    Dec 2007
    Posts
    22,013

    Default

    It sounds like it is the cross-atlantic link, if that is the only factor. And it's not unusual.

    Richard

  3. #3
    Join Date
    Sep 2011
    Posts
    1,920

    Default

    Hi
    Richard is correct, the only solution would be to get servers in Europe,

    If you decided to build up some servers in Europe,
    the problem you would face is getting the European clients to connect to the Europe server.

    There are a few ways of getting around this problem,

    1.) Have your website create different links for the European clients. (pointing them back to the European server)
    2.) Get a geographic load balancer to redirect clients to the closest server to them.

    Jason

  4. #4
    Join Date
    Aug 2011
    Posts
    6

    Default

    Hi,

    Thx for both replies.

    We are not sure its the cross-atlantic link, however we think its not a good factor and could explain alot of issues for the european users.

    We also receive complaints from user in a few km away from the datacenter. Users are experiencing the "Video stop and then rewind 2-3 seconds behind and then restart" .

    As I have read a few posts on this forum, some other suggestions were to increase the buffer length on the client side (we are using flowplayer).

    Now I've took some time to look at the VHost.xml file and I saw under the nextconnections xml tag the value SendBufferSize which is set to 60 000.

    I imagine that if I could increase the client buffertime I would also have to change the <MaximumSetBufferTime>60000</MaximumSetBufferTime> on the server side ?

    Increasing the buffer time would help thoses one with low BW or high lantency problems ? Again i could prolly be out of the track ...

    Thx for taking the time to read.

  5. #5
    Join Date
    Sep 2011
    Posts
    1,920

    Default

    Hi
    I think that to find out what the problem you are having is, ill to ask a few questions,
    Ok, first off do you know if its the same customers that are having these problems or are they totally different people every time?
    If its the same people then it could be that they have a slow connection and that may be a problem they will have to deal with.

    It could also be one or more of your servers, are the people that complain on the same servers?

    I personally wouldn't change things around just yet, if the service you provide works for 95% of your customers then changing settings sounds like a bad idea.

    Jason

Similar Threads

  1. High-availability cluster Active/Active for video-chat application
    By mmaron@mikogo.com in forum General Forum
    Replies: 1
    Last Post: 09-19-2014, 08:03 AM

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •