Reading other comments, I’ve realized that Plex will load moderately faster when an SSD is used. I realize that transcoding should NOT be done on the SSD because of excessive writing.
What I can’t seem to find: Does the whole Plex (PMS) install (/root directory) need to be on the SSD or is /config directory enough to benefit from preview speeds?
Someone correct me if I’m wrong here, but I assume the full PMS on SSD would only benefit initial startup of the application, not thumbnail preview/browsing - that is the responsibility of the metadata. Also, while PMS has regular write/read, metadata is mostly read; therefore, only moving metadata would put significantly less wear on an SSD.
And to clarify, the media files would be stored on standard spinners. The goal of this is not to affect speed of play time or buffering, but to make navigation faster.
Oh and this would apply to FreeNAS or linux setups where mouting with fstab is an option. Anyone have experience with this?
That was going to be my next question. If the default transcoder temp folder location is left blank (default), will Plex use /tmp in /root or /cache in Plex data folder or something else? And more importantly, will changing the Plex data folder location affect the default transcoder temp folder?
Also, I used an atypical install with the following (I believe that would define /config as the new Plex data folder though):
Just to be clear, changing the destination under Settings>Transcoder>“Transcoder temporary directory” will handle all excessive writing to SSD? Is there any way to check that this is working correctly?
By willfully causing transcoding during playback and then checking if the Cache/Transcode/Sessions subfolder inside the Plex data folder still starts to fill with transcoded data chunks. If it does, the rerouting was not successful.
Or, on the other hand, that the new Transcoder temp location is indeed starting to fill with data chunks during a transcoded playback session.
Keep in mind while mounting the transcoder temp location that you need to allow “execute” for it.
Plex also makes lots of little logging entries and does frequent database and statistics updates, but that’s exactly the stuff to keep on SSD if you want fast browsing performance.
The fear of “wearing out” SSDs is often overblown. It’s true, they have a lifespan, but it’s shared as tech voodoo. Just use your equipment and let your SSD be fast for you.
If you’re worried about it, you want to be using SMART and monitoring reallocated sectors. Unless that number is going up consistently, you’re fine.
What’s your drive? They almost all have published “write endurance” figures. The last tests I saw were years ago, and most drives exceeded those figures dramatically. Newer generations of drives have multiplied their write endurance, too.
Look at it this way - if it was a huge issue, somebody would have sued Apple for it.
For any future readers looking to do this:
The setting will use the /root directory and create/use a parent directory defined in the "Transcoder temporary directory and create a child directory called “Transcode”. Transcoding will occur in a child to that directory called “Sessions”. As @OttoKerner pointed out, data chunks will appear during an active transcode. All of this was accomplished successfully with a non-default support_path implemented at install, i.e. the Plex data folder is pointed to another drive.
You can create a new directory “/example” or use a pre-existing directory “/tmp” in the “Transcoder temporary directory” setting. I have NOT tested using the root directory or using a setting like “/”
I always suspected a bit of over-concern when it came to decent quality SSDs. If RAM can write infinitely, why shouldn’t SSDs.
In regard to Plex, would you suggest transcoding on the SSD? I can’t imagine any noticeable difference due to the size of the transcoder chucks.
In order to benefit from those on an SSD, do you know if that would require the entire PMS install to be on SSD or are those occurring in the Plex data folder?
I don’t avoid using the brakes on my car, just because they will wear out eventually. Or avoid driving at all, because the tires will wear out.
I’m not saying they don’t wear out. Just that it’s mostly not necessary to worry about, for most people, on a modern SSD.
For a single user, I imagine it’s somewhere between “no” and “placebo”. A single HDD can read the original file, write the transcoded file, and then read it again fast enough to stream to the user. And with luck that second read is cached anyway.
As you add more users it can help more and more. HDDs can easily read a single file at 100MB/s, but as soon as you’re doing read/write/read for multiple files, it looks more like random IO and total throughput goes down.
Some folks with bigger systems and many simultaneous transcodes add enough ram and use ramdisks for transcoding.
Now that you’ve familiar with the settings … hey, go test! Let us know!
