• Guest, The rules for the P & N subforum have been updated to prohibit "ad hominem" or personal attacks against other posters. See the full details in the post "Politics and News Rules & Guidelines."
  • Community Question: What makes a good motherboard?

Question How many video streams can a normal SATA slow 5400 rpm or 7200 rpm handle mostly on a Plex Server ?

sdifox

No Lifer
Sep 30, 2005
81,857
7,862
126
No realistic impact. A 8mbps steam is 1 megabyte per second. Duplex that for rw and it is 2 megabytes per sec. Even if you do 80mbps stream, it is 20 megabyte per second.

a 7200 rpm drive should be able to handle at least 150 megabyte write per second.
 
Last edited:

TheELF

Diamond Member
Dec 22, 2012
3,139
336
126
No realistic impact. A 8mbps steam is 1 megabyte per second. Duplex that for rw and it is 2 megabytes per sec. Even if you do 80mbps stream, it is 20 megabyte per second.

a 7200 rpm drive should be able to handle at least 150 megabyte write per second.
Doesn't take into account the seek rate for jumping from stream to stream...
My guess would be that more than one stream will start having at least minor issues.
 

sdifox

No Lifer
Sep 30, 2005
81,857
7,862
126
Doesn't take into account the seek rate for jumping from stream to stream...
My guess would be that more than one stream will start having at least minor issues.
you make it sound like hard drives never handled multiple access/write before. I am saying a hard drive can handle the rate plex needs even for multiple stream. 80mbps streams are ludicrous. You are gonna run out of cpu power before you can transcode enough to overwhelm a 7200rpm hdd


example, this is my Hitachi 2TB drive that is probably six year old
writing 16MiB block which is probably bigger than Plex write blocks.

Crystal Disk Mark Hitachi 2TB.PNG


don't put your scratch volume in the same drive as content drive is best practice.
 
Last edited:

TheELF

Diamond Member
Dec 22, 2012
3,139
336
126
You are gonna run out of cpu power before you can transcode enough to overwhelm a 7200rpm hdd
Isn't plex server for viewing movies?
I guess it could have a large enough buffer so that even an old drive can switch between the movies but what has transcoding to do with it?
I doubt that somebody working on serious video production would be so minimal in his questioning.
 

sdifox

No Lifer
Sep 30, 2005
81,857
7,862
126
Isn't plex server for viewing movies?
I guess it could have a large enough buffer so that even an old drive can switch between the movies but what has transcoding to do with it?
I doubt that somebody working on serious video production would be so minimal in his questioning.
Plex transcodes the video to the desired/supported resolution of your playback device. That is the only reason why write speed matters. If you are not transcoding, reading files is a cinch.
 

Billy Tallis

Senior member
Aug 4, 2015
237
99
101
example, this is my Hitachi 2TB drive that is probably six year old
writing 16MiB block which is probably bigger than Plex write blocks.
No, that's your Hitachi 2TB running a benchmark of 1MB and 4kB block sizes, with the benchmark duration limited to 16MB, which makes everything fit in cache and minimizes seek distance for anything that hits the real storage media, all of which makes it a meaningless test.

What you need to test are random writes of whatever block size Plex ends up writing, after taking into account caching done by the OS/filesystem (so run a disk trace and look at what kind of IO commands actually make it to the drive). The IO scheduler will probably try pretty hard not to totally interleave writes to different parts of the disk, and NCQ on the drive itself will help a tiny bit, but there's no doubt that hard drive throughput goes down fast when you start throwing in the occasional seek.
 

sdifox

No Lifer
Sep 30, 2005
81,857
7,862
126
No, that's your Hitachi 2TB running a benchmark of 1MB and 4kB block sizes, with the benchmark duration limited to 16MB, which makes everything fit in cache and minimizes seek distance for anything that hits the real storage media, all of which makes it a meaningless test.

What you need to test are random writes of whatever block size Plex ends up writing, after taking into account caching done by the OS/filesystem (so run a disk trace and look at what kind of IO commands actually make it to the drive). The IO scheduler will probably try pretty hard not to totally interleave writes to different parts of the disk, and NCQ on the drive itself will help a tiny bit, but there's no doubt that hard drive throughput goes down fast when you start throwing in the occasional seek.

16 megabyte per second ~= 128MBPS ... no one transcode to 128MBPS...

So the writes will be lower than 16 megabyte per second by stream.
Use RND4K Q32T16 as the benchmark numbers. I don't know anyone that runs Plex that transcode to the same hdd as the library drive. So the read write we are dealing with is writing the transcoded file to the scratch drive then serve that to the client.
 
Last edited:

ASK THE COMMUNITY