Planning for failover for streaming media service

No failover is possible for streams being delivered

You cannot set up a failover mechanism to handle streams that are in the process of being delivered to clients. Streaming connections are dropped if a problem occurs between the streaming media cache and the streaming server or between the streaming media cache and the clients. Ensuring that your network is sound should reduce the number of dropped connections.

The server streams the media stream to the cache, which then streams the media stream to the client. A connection between the server and streaming media cache remains open until the last client viewing the stream finishes viewing it. If a streaming media cache goes down, all connections that the streaming media cache had open between it and the streaming server are dropped. Users with connections that had been open must start playing the stream again.

If the origin server goes down, NetCache stops serving clients from that stream. New clients generate a new server connection.

Building redundancy into your streaming media deployment

You can build redundancy into your streaming media deployment in the following ways:

Add an extra streaming media cache.
You might want to plan for one streaming media cache in addition to the number you think that you need to handle the streaming traffic. That way, if one streaming media cache goes down, the remaining streaming media caches can still handle the volume of streaming traffic.
Use transparent proxying.
If one streaming media cache goes down, the L4 or L7 switch or WCCP router distributes new streaming media requests over the remaining streaming media caches. Note that this deployment does not prevent existing connections from being dropped.
- If you are using transparent proxying, configure your L4 or L7 switch or WCCP router to fail over to the Internet if all the streaming media caches go down.
- If you are using transparent proxying with an L4 or L7 switch, you can deploy a switch failover pair if you are concerned about the failure of your switch.
Use NetCache takeover if you are not using transparent proxying.
Deploy more than one streaming server with the same content. Then, if a cache miss occurs, the server does not become the point of failure.