First PMS build and VM gaming questions

After testing PMS on my old MacBook with good results on local network I decided to build my first dedicated server, but still have some questions and would appreciate some help.

My first plan for a build would be:

-OS: Ubuntu

-CPU Intel i3 8100T (focusing on HW transcoding)

-Mobo: any good h370 or z370

-RAM: any good 16gb DDR4 set (2x 8gb)

-PSU: Corsair CX550

-Storage: SSD (system), WD Red 8 or 10tb drives (data)

-Case: Fractal Design Define R6

It will be dealing with max 2-3 streams of 4K (local) and 1080p (local + remote) movies, mostly played on Infuse to avoid unnecessary transcoding.

My questions are:

  1. I notice many folks here use RAID options… my idea is to add more hard drives as library grows and would not storage any video that can’t be re downloaded if a drive fails. Is there any other advantage of using redundancy for this build Vs isolated partitions? (HDs are very expensive here in Brazil)

  2. I was also planning to build a mini-ITX gaming rig in the future and was reading some posts about using Windows 10 VM on the server setting a dedicated GPU for VM and Intel iGPU for system (Plex), if viable I would change CPU to 9700k and dedicate some cores for Plex. Anyone doing this around here can tell if worth it or would be better with separated systems? Looking for some ideas about power consumption as 24/7 PC (if the system would be able to operate on idle when VM is not in use) and other problems that I can expect.

  3. My idea of using SSD for system is to improve metadata access when library grows. But I’ve read some posts from Plex team that PMS constantly update files on the system and this could severely reduce the lifetime of SSD. Is this a real problem I need to be concern about or not?

  4. Most motherboards only have 6 SATA ports, if I decided to add a PCIe SATA card in the future, can this be a problem for Plex?

PS: Sorry for any English erros :slight_smile:

my opinions;

  1. the main advantage of redundancy is to avoid having to restore (or lose) your data to due a drive failure. So if you don’t care about all the time it takes having to re-rip or re-dl the lost media, then individual drives are indeed a viable option.

Some consider it is better to lose only 1 drive worth of data, instead of a whole array worth.

Either way, unless your data is backed up somewhere separate, if you lose it, your gonna have some kind of pain in re-acquiring it.

Also, no raid or redundancy is a replacement for backups. Backups mean multiple copies of your data on different media, and ideally in different locations (think of fire/flood/theft/etc).

  1. doing a vm is possible, but definitely is more work and more upkeep.

  2. theoretically, but modern SSD’s are designed for years of writes, I would expect it difficult to plex alone to wear out and SSD.

That said, I refer you to the first 3 rules of computers and media storage in particular… 1) backups, 2) backups, 3) backups of your backups.

  1. shouldn’t. plex doesn’t really care about the underlying hardware. All you do is point your library at a path, it can be local or a network share path.

Hope that helps, good luck.

Thanks. I know raid is not a backup solution and I keep all important stuff backup on external HDs.

I’ll look into some options of redundancy that don’t impact too much on storage capacity, but I’ve read some RAID options could cause problems with large (10tb) HDs, any thoughts on this?

The idea of using VM is to save space on apartment, but if it would be too problematic or keep sucking too much power I may end building a separated system in future. That was not on my plan at first as I’ve searched that Plex could face some problems with HW transcoding if a dedicated GPU is used, but with VM I can dedicate which GPUs and Cores will be available for system and for VM.

Well, raid is to help protect against disk failures (amongst other things), however as hard drives keep getting larger, it takes more time to rebuild/resync an array. During a rebuild/resync of raid, it puts a lot of disk load (reads/writes) on all of the drives in the array.

So, large disks = larger arrays + long times to resync + heavy IO load, increases the possibility of failure in other drives in the array. And if another drive fails during a resync or during a first disk failure, then all your data goes kaput.

Myself, I use raid 6, which is dual redundancy, so it would take 3 simultaneous disk failures to lose my data. However of course that reduces overall space and uses the extra disks for redundancy.

But I don’t want to have to restore 40+ terabytes and decades of mixed data that may or may not be replaceable, like old digitized vhs/cassette/records/pictures/etc.

I also have multiple nas devices, so I backup the main one to the others in smaller chunks.

Basically I started with one nas device, then upgraded to newer ones over time, and used the older ones as backup for the new one.

Having used raid 6 for a long time, I would never use raid 5 on large disks or large quantity of disks. The peace of mind is worth the extra cost and loss of usable storage, for me.

Raids and data storage is a whole topic with many dedicated websites and forums, I would recommend some researching if you want to learn more of the details and pro/cons of different setups.

Regarding power, if that is a concern I would recommend watching some of the videos @ https://www.youtube.com/channel/UCYnR50X5yj6a1acggumZzTg

They go into how much power can be saved going low power vs a big server.

And if power is a concern, you should consider how much power a gaming PC running idle full time with plex, versus a lower power plex dedicated machine.

it might be a small difference, might be large. you have to decide on the trade off of space consolidated on one machine vs power vs the expense of separate hardware.

One thing to keep in mind, regarding the hardware transcoding that is discussed in the videos, is if you expect to have 4k/hdr media and want to transcode it, you want either a quicksync version 6 compatible intel cpu, or a separate video card that is supported by plex hardware acceleration and supports 4k/hdr. Older video cards may support transcoding but not 4k/hdr.

Re downloading may be a pain, but hardware here in Brazil is about 6x more expensive due to currency and taxes. It’s cheaper to just re download everything, but thanks for the explanation on Raid.

If I could make the PMS and gaming in one PC would also save a lot of money and I think would do a better use of this big case, but I only see this working if somehow I could just disable the dedicated GPU and extra CPU cores when VM is not in use. Will keep looking into this.

I have never heard of a way to turn something like GPU let alone individual CPU cores off and on while the machine is running (or at all, really). However, the components should not use that much power at idle.

Have you decided upon a hypervisor yet? Keep in mind that most hypervisors don’t support plug-and-play passthrough to a VM. So in order to get USB plug-and-play, you‘d most likely need pass an entire USB controller through to a VM. You can get a PCI(-E) one and pass that to the VM to make it simple.

Check out Linus’ unraid gaming VM video… it’s pretty much the closest thing to VM gaming for beginners.

Didn’t look into that yet, just got this ideia of “dual” machine from some posts about unRaid and GPU passthrough.

That’s exactly the video that that gave me the idea of building everything in one PC :slight_smile:

This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.