• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

WD Blacks in 3 disk RAID-0?

R

RagnarKon

Junior Member
So I am looking into buying a third WD Black disk for my video editing rig. With 4K video becoming more popular, my current 2x 2TB RAID-0 disk is running out of room during the course of a project, so I am looking for a cheap alternative to add on another 1TB of space.

I know the "proper" way to do it would be to get some sort of direct attached storage device, but that isn't in the cards right now because of price. So my idea was to backup everything on my current stripped drive to my NAS, and then recreate the RAID-0 disk with the 3rd disk.

The problem is, I know the WD Black disks have no error recovery control (TLER? I think is what WD calls it). I've been told that it doesn't really matter for a simple two-disk RAID-0 or RAID-1... but I have no idea as to when something like ERC becomes important. Is it only important for RAID-5/10 environments? Or should I also be concerned about it on a 3-disk RAID-0 setup as well?

Note that data reliability isn't really that important in this case. I only use the RAID-0 as my "active" working drive, and I do nightly backups to my NAS. So at most I would lose 1 day of work. I just don't want to add in a 3rd disk if the RAID controller is going to continually drop disks from the array.
 
Last edited:
I don't think TLER is necessary if you're using software RAID, which for a RAID-0 on non-system files, should be just fine.

The easiest way to do it would be to grab a 4 TB external drive, copy the current 2x2TB RAID 0 to the external, back up to the NAS also (two backups!), physically install the 3rd disk, make a 3-way RAID-0 stripe of the three disks, then migrate the data from the 4 TB external to the newly created 3x2TB RAID-0.

RAID 0 is just JBOD of 2 disks.
Click to expand...
JBOD is not the same as RAID-0. RAID-0 actually stripes the files across all disks to improve performance but the volume size of the RAID-0 is limited to the capacity of the smallest disk times the number of disks. JBOD is allows you to get the full capacity of disks of different sizes, but doesn't offer any performance improvements.
 
Essence_of_War said:
I don't think TLER is necessary if you're using software RAID, which for a RAID-0 on non-system files, should be just fine.

The easiest way to do it would be to grab a 4 TB external drive, copy the current 2x2TB RAID 0 to the external, back up to the NAS also (two backups!), physically install the 3rd disk, make a 3-way RAID-0 stripe of the three disks, then migrate the data from the 4 TB external to the newly created 3x2TB RAID-0.


JBOD is not the same as RAID-0. RAID-0 actually stripes the files across all disks to improve performance but the volume size of the RAID-0 is limited to the capacity of the smallest disk times the number of disks. JBOD is allows you to get the full capacity of disks of different sizes, but doesn't offer any performance improvements.
Click to expand...

Actually it is. JBOD SPAN -> is linear. JBOD Stripe is Non-RAID stripping. Windows home server used JBOD also with its disk pool by keeps 2 or 3 copies of a raw file spread on disks. "Just a bunch of disks" is what non-"RAID" 0 is.
 
Last edited:
Atreidin said:
JBOD is definitely not RAID-0.
Click to expand...

"RAID 0" certainly isn't RAID. It is a JBOD strip set. "RAID-0" was the name it got because it used RAID like striping techniques but was RAID-Nothing. IE the "Redundant" part doesn't apply. The only benefit it has is that SNIA assigned it a code to be recognized by controllers. ZFS no parity strip is another example of RAID like techniques with "just a bunch of disks" however it is not referred to as RAID 0 officially since it is not a SNIA recognized standard.

It may be a bit nit picky but that doesn't make the statement incorrect.
 
Last edited:
imagoon said:
"RAID 0" certainly isn't RAID. It is a JBOD strip set. "RAID-0" was the name it got because it used RAID like striping techniques but was RAID-Nothing. IE the "Redundant" part doesn't apply. The only benefit it has is that SNIA assigned it a code to be recognized by controllers. ZFS no parity strip is another example of RAID like techniques with "just a bunch of disks."
Click to expand...

I get your point about not having redundancy; however, JBOD and RAID-0 have different meanings in common usage. A striped set of drives typically isn't referred to as JBOD despite not having redundancy.

OP - You won't have any issues as you don't need to worry about TLER for RAID-0. As you don't have redundant data, you don't want the drive to stop trying to recover data (you have no other option).
 
Wuzup101 said:
I get your point about not having redundancy; however, JBOD and RAID-0 have different meanings in common usage. A striped set of drives typically isn't referred to as JBOD despite not having redundancy.

OP - You won't have any issues as you don't need to worry about TLER for RAID-0. As you don't have redundant data, you don't want the drive to stop trying to recover data (you have no other option).
Click to expand...

I would argue the "common usage" point, clarifying that I mean in storage in general. In the enthusiast arena sure I would agree with JBOD = spanning disks linearly and RAID-0 stripped disks. JBOD is a class just like RAID is a class. RAID-0 is a specific configuration of a JBOD of disks. Once you get in to "real" storage the concepts shift significantly. Using VNX as an example, you can create a pool of disks (Fancy name for JBOD.) In that pool you can then strip the blocks with no parity, single parity, double parity like Raid 0, 5 and 6. However the blocks are not constrained to the disks so double parity "Block 1" can be on disks 0-6 and then "block 2" is on disks 7-13. Going even farther there is no reason why a double parity and single parity block can't share the same disk pool (the often do.) You can use these techniques to get RAID 50 and 60 performance out of the pools without the extra loss of a dedicated resilient set. In addition the pool isn't constrained by the type or speeds of the disks and can use intelligence to move stale data to slow disks in the pool and high load data to the high speed disks in the pool.

These are not recognized "raid levels" from SNIA. It is actually a JBOD using RAID like techniques (XOR etc.) This is the reason RAID isn't dead (like so many Anandtech posters claim.) It is just evolving.
 
Last edited:
IMO... I personally prefer the term "RAID-0" over JBOD strip or span or whatever you wanna call it. Although I can see that RAID-0 certainly doesn't have the "Redundancy" part. Should just called it a AID-0 😉

So as I understand it, TLER is just a time-limit before the disk gives up on reading the data, correct? If you have a RAID-5, this isn't a big deal because you can just XOR the bits to recover the lost data. But if you have a RAID-0... you HAVE to find the data because it isn't stored anywhere else. Therefore TLER is irrelevant. So essentially, for a stripped set, it doesn't matter if TLER is enabled or not—because if the data isn't there you aint gotta get it from anywhere else. Am I on the right track?

And my RAID is a hardware RAID (via the chipset) for performance reasons. I'm not sure if that changes anything over having a software RAID. The whole purpose of the RAID-0 is to keep performance up while editing 4K video... and while SSDs would obviously be the best choice... high capacity SSDs also aren't exactly cheap.
 
Last edited:
That's correct. If you were running something like a RAID 5 setup, you would want the drive to stop trying to recover so that the controller could just fix the issue from the information on the parity disk. RAID 0... you have no option but to have the disk keep trying.
 
Small update, just in case anyone happens to Google this thread in the future. The setup works fine, haven't any any problems what so ever.

But the speed improvement from 2 drives to 3 drives is MUCH less than from 1 drive to 2 drives.
1->2 saw nearly a 70% speed increase.
2->3 saw only about a 35% increase.
 
is it fair to say the red drives are more meant for raid arrays and all the other wd drives are meant to be run normally?
 
If your raid setup is half decent you should be able to just grow the existing array without wiping anything. Keep in mind that raid 0 has no redundancy though... But if this is for video editing I imagine this is for temp storage of raw files till you do the final render? You probably still want to backup that stuff though in case you need to change something and render again.

I would not bother with the red drives unless you are using hardware raid.

I have 8 WD blacks in raid 5 in my storage array right now, they work pretty well.
 
Last time I built a "big" RAID array, it was a hardware RAID5. I merely used four 320GB Seagate desktop drives without the TLER features. Never had a problem with it; never had to replace a disk in the array.

I'm beginning to think the performance attractions of RAID0 no longer hold the value they used to have in any enthusiast system. If you boot from a separate SSD, you would think that simple file access performance would not figure that much by using the HDDs as JBOD or some sort of drive-pool setup.

But yes -- you could simply add a third disk to the RAID0-of-two and force a RAID rebuild.
 
Red Squirrel said:
If your raid setup is half decent you should be able to just grow the existing array without wiping anything. Keep in mind that raid 0 has no redundancy though... But if this is for video editing I imagine this is for temp storage of raw files till you do the final render? You probably still want to backup that stuff though in case you need to change something and render again.

I would not bother with the red drives unless you are using hardware raid.

I have 8 WD blacks in raid 5 in my storage array right now, they work pretty well.
Click to expand...

Pretty much this.

TLER just prevents the HDDS from dropping out of the array too quickly but that only happens with hardware RAID. If speed is your ultimate goal and data integrity is not important, I'm not sure I would even bother with Blacks. I would probably just get some 7200RPM Blues or Seagate Barracudas and toss them in there.
 
But yes -- you could simply add a third disk to the RAID0-of-two and force a RAID rebuild.
Click to expand...

Pray tell, what redundancy is your RAID-0 rebuilding from? 😵

I don't think what you're describing is a rebuild, it sounds like more of a restore-from-backup.
 
Essence_of_War said:
Pray tell, what redundancy is your RAID-0 rebuilding from? 😵

I don't think what you're describing is a rebuild, it sounds like more of a restore-from-backup.
Click to expand...

It seemed to me that any array I've had with striping had a feature so that you could just add another drive and expand the array. In most cases, I would just start with so many disks and then add another later.
 
BonzaiDuck said:
It seemed to me that any array I've had with striping had a feature so that you could just add another drive and expand the array. In most cases, I would just start with so many disks and then add another later.
Click to expand...
Just a word of caution, be very mindful when using this feature! It'll probably work a lot better with identical drives or similar capacities but a guy at an office tried to expand his Raptor 300GB with a WD Black 2GB in a JBOD which theoretically should have worked... corrupted both HDD's and the guy tried to blame me for his PC being offline for so long as I tried to recover data. Mr. big shot AutoCAD guy should have called us to install the HDD for him.

Anywho, moral is you need good backups to mess with something like this and if you have good backups you don't need to rebuild. Fresh installs are always better.
 
PliotronX said:
Just a word of caution, be very mindful when using this feature! It'll probably work a lot better with identical drives or similar capacities but a guy at an office tried to expand his Raptor 300GB with a WD Black 2GB in a JBOD which theoretically should have worked... corrupted both HDD's and the guy tried to blame me for his PC being offline for so long as I tried to recover data. Mr. big shot AutoCAD guy should have called us to install the HDD for him.

Anywho, moral is you need good backups to mess with something like this and if you have good backups you don't need to rebuild. Fresh installs are always better.
Click to expand...

Those are my favorite people. "I did something stupid and since you weren't here to stop me... IT IS ALL YOUR FAULT."
 
PliotronX said:
Just a word of caution, be very mindful when using this feature! It'll probably work a lot better with identical drives or similar capacities but a guy at an office tried to expand his Raptor 300GB with a WD Black 2GB in a JBOD which theoretically should have worked... corrupted both HDD's and the guy tried to blame me for his PC being offline for so long as I tried to recover data. Mr. big shot AutoCAD guy should have called us to install the HDD for him.

Anywho, moral is you need good backups to mess with something like this and if you have good backups you don't need to rebuild. Fresh installs are always better.
Click to expand...

A 2 GB Black drive?!! I don't even think the model line included a drive that small, so you're probably mistaken or there was a typing error. That being said, you're spot-on over that issue. I'd always expected to use the identical model drive and size to expand an array. I always anticipated, when push came to shove and I needed to replace a drive, I'd at least have one of similar size. But I always ordered spares when I build a RAID.

That's probably one reason I'm not all hot and bothered about RAID0 and haven't been for a few years. No less with RAID5. If you were going to build a 4-disk array, you'd at least be wise to buy a fifth disk to anticipate one of the four going bad. If you waited too long to acquire such a disk, or didn't make an effort to do it in the first place, the specific replacement drive wouldn't be available. OF course, you might be able to SHRINK an array to accommodate that sort of problem, but less feasible.
 
imagoon said:
Those are my favorite people. "I did something stupid and since you weren't here to stop me... IT IS ALL YOUR FAULT."
Click to expand...
Yeah, he seriously tried to tell us that I was costing his company money because he didn't have his AutoCAD like same day pronto.

BonzaiDuck said:
A 2 GB Black drive?!! I don't even think the model line included a drive that small, so you're probably mistaken or there was a typing error. That being said, you're spot-on over that issue. I'd always expected to use the identical model drive and size to expand an array. I always anticipated, when push came to shove and I needed to replace a drive, I'd at least have one of similar size. But I always ordered spares when I build a RAID.

That's probably one reason I'm not all hot and bothered about RAID0 and haven't been for a few years. No less with RAID5. If you were going to build a 4-disk array, you'd at least be wise to buy a fifth disk to anticipate one of the four going bad. If you waited too long to acquire such a disk, or didn't make an effort to do it in the first place, the specific replacement drive wouldn't be available. OF course, you might be able to SHRINK an array to accommodate that sort of problem, but less feasible.
Click to expand...
It was for sure a 2TB, the Black line has been long running I actually have a 640GB Black in a Q6600 system. Yeah, RAID in general needs some forethought even RAID-5 I mean you've got one parity disk and then if another goes offline before replacement and lengthy rebuild process the entire array is toast and not to mention mucho overhead. If I were to do it right, it'd be 10 with image backups to another system.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top