Direct Play Buffering Functionality?

[Luke 37252]

Let's please try to keep this thread on topic about buffering. Thanks guys.

[FrostByte 5061]

https://github.com/MediaBrowser/Wiki/wiki/Direct-Play,-Direct-Stream-&-Transcoding 

 

Does that put this to rest? 

Cayars woz 'ere. 

Makes things so much clearer, thanks

[slicedmass 2]

Okay for clarity's sake. When Emby is serving up the video (instead of direct file SMB or what have you access) untouched or just remuxed on the fly. It would be nice if we could set a cache value so that Emby will keep the next xxxMB in memory with a rolling cache (dump the old already watched portion for the next). Something like if you set cache to 200MB per user or whatever. Emby will break it up in to 10 pieces 20MB each. The client requests data, Emby server pulls it from the memory cache. Once the Emby server passes a full 20MB chunk, it is purged from the cache and Emby server hits the Disks for the next 20MB chunk. Rinse repeat. If the client seeks outside of the 200MB worth of cache, its all purged and we start again. This would greatly reduce IOP hit on the disks and you would be more likely to bottleneck ur network before your spinning disks, this also puts the latency almost entirely on the network and avoids any Disk latencies which grows as the user count grows.

[Carlo 4331]

Your operating system should be doing this type of thing already.  All modern OSes will use memory to read ahead of the file being requested.