Plex Server

[Br3tt96]

So I wanna transition my old i5 plex server with single WD red 6tb into my current i7-4770k with a 4x8-10 shucked drive setup. I'm not sure the best way to go about setting up raid or even using unraid. I'd like something I could possibly have some redundancy cause JBOD backing up is just annoying. Like some sort of mirrored drives, so maybe Raid 1 or 5?

 

Got any suggestions or advice on this?

[JTPOTATO]

So my Plex setup is 5 x 3TB in RAID5, gives plenty of throughput, without using too many disks for redundancy, but still gives that redundancy

[Senzelian]

7 minutes ago, Br3tt96 said:

4x8-10 shucked drive setup

Do you mean 4x 8TB drives or what is it that you mean?

The amount of drives and the specific use case determine the RAID level.

 

I'm currently running Plex on my Synology NAS with 4x 1TB drives in RAID 5. Works well enough, but only one drive can fail.

The more drives of one type you have, the more likely it is, that more than just one drive fails.

[TehDwonz]

9 minutes ago, Br3tt96 said:

So I wanna transition my old i5 plex server with single WD red 6tb into my current i7-4770k with a 4x8-10 shucked drive setup. I'm not sure the best way to go about setting up raid or even using unraid. I'd like something I could possibly have some redundancy cause JBOD backing up is just annoying. Like some sort of mirrored drives, so maybe Raid 1 or 5?

 

Got any suggestions or advice on this?

I use Drivepool (and Scanner for fault detection) from https://stablebit.com/ because Plex really doesn't need the speed of RAID5/6. Besides, error rates on large drives lead to issues with "bit rot" in RAID5. I'd rather not have to re-rip my movies, so they are all duplicated.

[Senzelian]

5 minutes ago, TehDwonz said:

Besides, error rates on large drives lead to issues with "bit rot" in RAID5.

Got some good sources about that issue? I'd love to read more about it!

[Br3tt96]

13 minutes ago, JTPOTATO said:

So my Plex setup is 5 x 3TB in RAID5, gives plenty of throughput, without using too many disks for redundancy, but still gives that redundancy

Have you ever had to rebuild a drive?

6 minutes ago, Senzelian said:

Do you mean 4x 8TB drives or what is it that you mean?

The amount of drives and the specific use case determine the RAID level.

 

I'm currently running Plex on my Synology NAS with 4x 1TB drives in RAID 5. Works well enough, but only one drive can fail.

The more drives of one type you have, the more likely it is, that more than just one drive fails.

I meant I was gonna go with 4x8TB or possibly 4x10TB depending on sales during black Friday. Unfortunately, could not pick any up on the Prime day sale. 

4 minutes ago, TehDwonz said:

I use Drivepool (and Scanner for fault detection) from https://stablebit.com/ because Plex really doesn't need the speed of RAID5/6. Besides, error rates on large drives lead to issues with "bit rot" in RAID5. I'd rather not have to re-rip my movies, so they are all duplicated.

Yeah I'll look into this. I really don't wanna have to re-rip stuff since my collection has grown exponentially over 2 years.

[Electronics Wizardy]

8 minutes ago, TehDwonz said:

I use Drivepool (and Scanner for fault detection) from https://stablebit.com/ because Plex really doesn't need the speed of RAID5/6. Besides, error rates on large drives lead to issues with "bit rot" in RAID5. I'd rather not have to re-rip my movies, so they are all duplicated.

THis is a pretty over hyped issues, the real error rates on hdds, are much lower than the rated rates on drive sheets.

 

Also drive pool won't really help you here compared to traditional raid, if you have a read error during a rebuilt, you still won't get the data back.

 

Just make sure to have backups and you won't lose data.

[TehDwonz]

11 hours ago, Senzelian said:

Got some good sources about that issue? I'd love to read more about it!

Wrote in a rush earlier, and left out part of the sentence... that reads as though I am saying error rates are worse on large drives - they aren't. What I meant to say, was that error rates have not improved in the same way that capacity has over the years. You can see arguments on various sides of this if you Google "hard disk bit rot"

 

12 hours ago, Electronics Wizardy said:

THis is a pretty over hyped issues, the real error rates on hdds, are much lower than the rated rates on drive sheets.

 

Also drive pool won't really help you here compared to traditional raid, if you have a read error during a rebuilt, you still won't get the data back.

 

Just make sure to have backups and you won't lose data.

Yeah, you're right, and I rushed my reply, sorry. It's possible to combine a drivepool with ReFS instead of NTFS if that's available to you. Others use software SnapRAID set up to function as a data integrity check (scrubbing), on top of the duplication - but I think that over-complicates things.
 

My bigger reason to not use RAID, is that I also prefer my data to be easily accessible during recovery/repairs or if the RAID controller dies. With RAID, you can't just take a drive and plug it in elsewhere to access files. With Drivepool, it's just a normal NTFS/ReFS drive, accessible without special software. Losing a single disk under RAID5 vs DP, both are still usable arrays - RAID5 will need to re-build, which is a pretty drive-thrashing process for all the disks. DP makes the duplicates as a background process, giving way to file access requests, but only involves disks in pairs at any given time. Both take about the same time for a given amount of data, but the DP "drive" will work at full speed if needed (obv. depending on you using a good SAS/SATA controller).
Now take losing 2 or more drives at once (and DP set to 2x dupe only). In both cases, you lose some data. However, the RAID5 array breaks until you rebuild - DP keeps working, for data that was not lost entirely, at full speed. Less downtime as a result, as it can be left operational while failed disks are replaced, and no re-build is needed. The replacement disks just get treated as adding new disks to the setup, and data is duplicated and optionally balanced across drives in the background. You can then restore any lost data from That Actual Backup You Totally Have(tm).

 

Drive failure (false positive or not) on consumer models is still around 10% after 3-4 years always-on use - though you'll find this value contested a fair bit either way. Backblaze says higher, others say lower, but sample sizes are in contention on most claims either way. Personally, I plan for 10% over 3 years for consumer drives, and that includes WD Red (non Pro) and Seagate NAS, as well as non NAS desktop models. That goes up after year 4 - with failure becoming essentially inevitable as time goes on.

 

You'd have to work out the costs for your own personal setup, weighing up initial disk cost to buy (Pro vs non-pro), time taken to replace on failure, and time taken to restore full operational function. Don't forget the disk controller dying too - and it does happen. For smaller arrays, maybe RAID might be more convenient, but once you get to 8-10 disks, I find the software duplication option to be more cost effective.

 

Lastly, on the topic of file access and speeds, DP offers a "psuedo-striping" for read operations of duplicated files. Say a file is being read (a movie MKV is being streamed on Plex), and you or someone else wants to read that file too. DP will cause the second request to be read from the disk that file is duplicated on, in parallel. So while you won't get the higher combined speeds you would from RAID5, parallel access allows multiple full speed reads of the same data, and this scales with your duplication factor - just like an Ethernet LAGG setup of 4 ports gives you 4 lots of 1Gbps rather than 4Gbps.

 

TLDR: RAID recovery is more bother than simple file duplication, and the single-file speed advantage of RAID is not needed for Plex, even with 4K "raw" BD rips.

If you do go RAID, make sure you get a decent battery-backed controller, and don't use your cheapo motherboard RAID unless you like misery.

[Br3tt96]

5 hours ago, TehDwonz said:

Wrote in a rush earlier, and left out part of the sentence... that reads as though I am saying error rates are worse on large drives - they aren't. What I meant to say, was that error rates have not improved in the same way that capacity has over the years. You can see arguments on various sides of this if you Google "hard disk bit rot"

 

Yeah, you're right, and I rushed my reply, sorry. It's possible to combine a drivepool with ReFS instead of NTFS if that's available to you. Others use software SnapRAID set up to function as a data integrity check (scrubbing), on top of the duplication - but I think that over-complicates things.
 

My bigger reason to not use RAID, is that I also prefer my data to be easily accessible during recovery/repairs or if the RAID controller dies. With RAID, you can't just take a drive and plug it in elsewhere to access files. With Drivepool, it's just a normal NTFS/ReFS drive, accessible without special software. Losing a single disk under RAID5 vs DP, both are still usable arrays - RAID5 will need to re-build, which is a pretty drive-thrashing process for all the disks. DP makes the duplicates as a background process, giving way to file access requests, but only involves disks in pairs at any given time. Both take about the same time for a given amount of data, but the DP "drive" will work at full speed if needed (obv. depending on you using a good SAS/SATA controller).
Now take losing 2 or more drives at once (and DP set to 2x dupe only). In both cases, you lose some data. However, the RAID5 array breaks until you rebuild - DP keeps working, for data that was not lost entirely, at full speed. Less downtime as a result, as it can be left operational while failed disks are replaced, and no re-build is needed. The replacement disks just get treated as adding new disks to the setup, and data is duplicated and optionally balanced across drives in the background. You can then restore any lost data from That Actual Backup You Totally Have(tm).

 

Drive failure (false positive or not) on consumer models is still around 10% after 3-4 years always-on use - though you'll find this value contested a fair bit either way. Backblaze says higher, others say lower, but sample sizes are in contention on most claims either way. Personally, I plan for 10% over 3 years for consumer drives, and that includes WD Red (non Pro) and Seagate NAS, as well as non NAS desktop models. That goes up after year 4 - with failure becoming essentially inevitable as time goes on.

 

You'd have to work out the costs for your own personal setup, weighing up initial disk cost to buy (Pro vs non-pro), time taken to replace on failure, and time taken to restore full operational function. Don't forget the disk controller dying too - and it does happen. For smaller arrays, maybe RAID might be more convenient, but once you get to 8-10 disks, I find the software duplication option to be more cost effective.

 

Lastly, on the topic of file access and speeds, DP offers a "psuedo-striping" for read operations of duplicated files. Say a file is being read (a movie MKV is being streamed on Plex), and you or someone else wants to read that file too. DP will cause the second request to be read from the disk that file is duplicated on, in parallel. So while you won't get the higher combined speeds you would from RAID5, parallel access allows multiple full speed reads of the same data, and this scales with your duplication factor - just like an Ethernet LAGG setup of 4 ports gives you 4 lots of 1Gbps rather than 4Gbps.

 

TLDR: RAID recovery is more bother than simple file duplication, and the single-file speed advantage of RAID is not needed for Plex, even with 4K "raw" BD rips.

If you do go RAID, make sure you get a decent battery-backed controller, and don't use your cheapo motherboard RAID unless you like misery.

I pretty much wanna do a mirror to mirror copy, so would it be better to just run JBOD? Make a movies drive, movies backup, tv show drive, and tv show back up and just move files to both drives as I update with new media?

[Electronics Wizardy]

5 hours ago, Br3tt96 said:

I pretty much wanna do a mirror to mirror copy, so would it be better to just run JBOD? Make a movies drive, movies backup, tv show drive, and tv show back up and just move files to both drives as I update with new media?

How many tb of movies do you have?

 

Id setup a raid array on the main system to store all the files in one location, then setup a backup to anouther location incase something happenes to that array.

[Br3tt96]

45 minutes ago, Electronics Wizardy said:

How many tb of movies do you have?

 

Id setup a raid array on the main system to store all the files in one location, then setup a backup to anouther location incase something happenes to that array.

So I have my plex on a i5-3570k system with a single 6tb wd red. It's about %65 full atm. I'd like to shuck some drives for more storage and move it to the i7-4770k build in my signature. Trying to figure out the best way to do so

[Electronics Wizardy]

4 hours ago, Br3tt96 said:

So I have my plex on a i5-3570k system with a single 6tb wd red. It's about %65 full atm. I'd like to shuck some drives for more storage and move it to the i7-4770k build in my signature. Trying to figure out the best way to do so

Id probably do a raid 5 of 4 of those 8-16tb drives then. Then setup backups in case of a raid failure.

 

Can go with something like unraid if you want easy expansion, really depends on your goals

[Br3tt96]

12 hours ago, Electronics Wizardy said:

Id probably do a raid 5 of 4 of those 8-16tb drives then. Then setup backups in case of a raid failure.

 

Can go with something like unraid if you want easy expansion, really depends on your goals

Yeah I was thinking unraid might be the best option. I also have unlimited Google Drive storage as well for free, so I guess you could say that's my backup