Skip to main content
M
Michael_Kennedy__12345_
Inspiring
June 10, 2010
Question

Stream Failover on Encoder Failure

  • June 10, 2010
  • 1 reply
  • 1412 views

Hi,

I am looking for advise on setting up live encoders to provide redundancy. I am using FMLE 3.1, but I presume that any server side solution would support all encoders. A basic version of the problem would be where I have one server and two encoders (with the same video source over SDI). If one encoder fails I want to be able to deliver the feed from the second encoder, without having client side logic. Is this possible?

I would welcome a discussion on ways to provide improved resiliency in the live streaming environment. It is amazing that no out-of-box solution is available to provide such basic functionality.

Michael

BTW: I asked a similar question some months ago but got no answers. Apologies for bringing it up again, but this must be a fairly common problem, as anybody using FMS and FMLE will need some solution. I have tried the livestreamcast application, but had little success. It also appears to be for a much bigger scale than I have, requiring an all origin environment.

    This topic has been closed for replies.

    1 reply

    Asa_-_FMS
    Adobe Employee
    Asa_-_FMS
    Adobe Employee
    June 11, 2010

    Hi Mike,

    So that I can direct your question - (which falls a bit on the fence) is the issue in sorting how to get FMLE or your encoder of choice to publish more than one stream at once?  Or that's already working and you're looking for how FMS handles the case?

    Asa

    M
    Michael_Kennedy__12345_Author
    Inspiring
    June 12, 2010

    Hi Asa,

    The issue is getting TWO FMLE encoders to publish the same stream to ONE flash server at the same time, so that the failure of an encoder does not cause the stream to fail.

    I am using the default "live" application on FMIS. By encoder I am refering to an individual server running one instance of FMLE. Two encoders refers to two independent servers, in separate physical locations, running FMLE and pushing the same stream to one server(rtmp://mysserver/live/mylivestream) . The end goal is a full mesh of Origin servers and Encoders, so that my end users do not notice any stream failure due to the failure of any individual component. Right now, I need to failover manually if an encoder dies, which could result in significant disruption depending on the timing of the failure. I have a reasonable solution in place for FMIS origin server redundancy, but as I said in the first post, I would welcome ideas from the forum on how an overall solution is best put together.

    Thanks,

    Michael

    S
    SE_0208
    Participating Frequently
    June 14, 2010

    Hi Micheal,

    There are ways you can achieve redundancy but I do not think you can achieve without extra client-side code. Just take your case of two FMLE publishes same source to the same server. Now two instances can take same source but it cannot publish with same stream name and if it is doing so then it has to be to different application or different instance. Now this calls for inherent code logic where client has to issues play for different stream name or same stream name over different netconnection. So i suppsoe you would have to have extra failover logic in client.

    If you are ok with extra client logic, we can further on discuss how can we achieve better failover.

    Terms & ConditionsAccessibility statement
    Using the community Terms of use Privacy Cookie preferences Do not sell or share my personal information ad choicesAdChoices