NETGEAR is aware of a growing number of phone and online scams. To learn how to stay safe click here.
Forum Discussion
Solved
What protection does XRaid offer for large disk arrays?
I’ve been running my 314 for 3 years now with a mixture of disks Disk 1 – Seagate 2TB (ST32000542AS 24k+ hours) Disk 2 Seagate 3TB (ST3000DM001 18k+ hours, the one that has high failure rates) ...
Retired_Member wrote:
Instead, in your situation I would either go with
1) 4 WD Red 3TB or
2) 3 8TB (WD or HGST preferred)
As has been pointed out, XRAID will let you use 2x8TB+3TB with no issue.
Just to comment on the economics, using current Amazon US pricing:
WD30EFRX: $110.00
WD80EFZX: $265.00
4x3TB costs $330 (since he is reusing the WD30EFRX he already has), that gives him 6 TB more storage than the existing WD30EFRX has by itself. So that is $55 per TB gained.
2x8TB+3TB costs $530, and gives 8 TB more storage than the existing 3 TB drive. That is $66.25 per TB gained.
So it is the case that 4x3TB is more economical in the short run. But later on, you will pay a much higher price to expand it again.
For example,
- adding 5 TB to the 4x3TB system will cost $530 (upgrading two drives to WD80EFZX). That is $106 per TB gained.
- adding 8 TB to the 2x8TB+3TB system costs only $265 (adding a single WD80EFRX) That's only $33 per TB gained.
hp532 wrote:
I’ve been reading up on Raid5 and wonder what protection XRaid (which I understand to be Raid5 + volume expansion) offers when used with large capacity disks? Most places suggest that given disks have an error rate of 1 in 10e14 you’d expect a 11TB array rebuild after a disk failure to fail with around 90% probability (11/12.5).
Those are the "death of RAID" sites I think.
But lots of people here have single-redundancy arrays that are bigger than 12 TB, and they have resynced them successfully (often many times). Every time you do a scrub you are reading all the sectors, and I do that every three months on my 15 TB, 16 TB, and 18 TB RAID-5 volumes with no problems. The math says that just won't work.
So it's clear that you can't just blindly apply the math. The error rate is spec'd at < 10e14, and the "less than" clearly is important. Drives are much more reliable than that spec suggests.
Raid5 does need at least 3 drives to create the redundancy protecting you against one of the drives getting faulty.
For consistency reasons I would not group one 3TB with two 8TB drives.
Instead, in your situation I would either go with
1) 4 WD Red 3TB or
2) 3 8TB (WD or HGST preferred)
Case1)
- You only would get about 9TB (3x3TB data and 1x3TB for redundancy) and have 4 drives whereof 1 is allowed to get faulty
+ More economic solution as the 3 additional drives are significantly cheaper, than 3 8TB drives of whatever brand
Case2)
- More expensive solution
+ You would get about 16TB (2x8TB data and 1x8TB for redundancy) and have 3 drives whereof 1 is allowed to get faulty
+ You could add another 8TB drive to end up with about 24TB (3x8TB data and 1x8TB for redundancy) and have 4 drives whereof 1 is allowed to get faulty
+ Solution just has more flexibility depending on your future data needs
My preferred option would be 2), if cost is not a critical issue. Otherwise would go with 1).
Retired_Member wrote:
Instead, in your situation I would either go with
1) 4 WD Red 3TB or
2) 3 8TB (WD or HGST preferred)
As has been pointed out, XRAID will let you use 2x8TB+3TB with no issue.
Just to comment on the economics, using current Amazon US pricing:
WD30EFRX: $110.00
WD80EFZX: $265.00
4x3TB costs $330 (since he is reusing the WD30EFRX he already has), that gives him 6 TB more storage than the existing WD30EFRX has by itself. So that is $55 per TB gained.
2x8TB+3TB costs $530, and gives 8 TB more storage than the existing 3 TB drive. That is $66.25 per TB gained.
So it is the case that 4x3TB is more economical in the short run. But later on, you will pay a much higher price to expand it again.
For example,
- adding 5 TB to the 4x3TB system will cost $530 (upgrading two drives to WD80EFZX). That is $106 per TB gained.
- adding 8 TB to the 2x8TB+3TB system costs only $265 (adding a single WD80EFRX) That's only $33 per TB gained.
hp532 wrote:
I’ve been reading up on Raid5 and wonder what protection XRaid (which I understand to be Raid5 + volume expansion) offers when used with large capacity disks? Most places suggest that given disks have an error rate of 1 in 10e14 you’d expect a 11TB array rebuild after a disk failure to fail with around 90% probability (11/12.5).
Those are the "death of RAID" sites I think.
But lots of people here have single-redundancy arrays that are bigger than 12 TB, and they have resynced them successfully (often many times). Every time you do a scrub you are reading all the sectors, and I do that every three months on my 15 TB, 16 TB, and 18 TB RAID-5 volumes with no problems. The math says that just won't work.
So it's clear that you can't just blindly apply the math. The error rate is spec'd at < 10e14, and the "less than" clearly is important. Drives are much more reliable than that spec suggests.
I am a heavy believer in dual redundancy especially as they come out with ever larger disks.
Consider these factors;
when you rebuild/resync a raid 5 device, you are putting a heavy load on all of your drives.
the larger the array is, the longer it takes for a rebuild to finish
so you are putting a heavy load, for a longer period of time, the bigger your array is
if your array is rebuilding because of a failed disk, and your other disks have similar hours on them, you are now putting a heavy load, for a long period of time, on disks that have a lot of hours on them already.
with the 6 and 8 bay devices, raid6/dual redundancy is the default, but if you started out with only a few disks you might still be in raid 5/single redundancy mode.
Either way, with only a single redunancy, if anything bad happens during a rebuild, you are likely screwed plain and simple. this forum and every other nas forum are littered with the tears of those who assumed that raid 5 would protect them and did not have a separate backup of their data.
neither RAID or a NAS is backup by itself. neither will protect you from theft, fire, flood, or malicious users.
a backup means a completely separate copy of your data, ideally multiple copies, in multiple different places.
raid6/dual redundancy will better help keep your data safer during rebuilds/resyncs, if a drive fails during a raid 6 rebuild, you will still have your data intact.
Of course if you have really bad luck and multiple drives fail, again you are likely screwed without a separate backup.
in summary, neither raid, nor nas is a replacement for backups, backups, and more backups.
RAID modes aside, you can't extrapolate RAID breakdown from the URE specs - 1 in 10e14 is clearly very conservative. I suspect that's to motivate people to buy enterprise-class drives. But that doesn't mean that multiple drive failures don't occur - they do, and lots of people here have been hurt by them.
TeknoJnky wrote:
I am a heavy believer in dual redundancy especially as they come out with ever larger disks.
That's a reasonable approach, as long as people don't use that as a subsitute for backup (which of course you don't).
I use single redundancy myself, but I maintain two backups on other ReadyNAS (the three volumes I mentioned above are my primary and two backup NAS).
Thanks all for you help. I think I will go with the 3TB, 8TB, 8TB option as this allows for future expansion. I've already purchased a 8TB Seagate Iron Wolf. I'll probably go with a WD Red for the second 8TB drive. I prefer HGST, but the HGST 8TB NAS drive is not on the compatibility list. Interestingly, at least on paper these Iron Wolf drives have much better specs. The 6TB and above have an error rate of 10e15 versus the 10e14 of the Red.
Those are the "death of RAID" sites I think.
But lots of people here have single-redundancy arrays that are bigger than 12 TB, and they have resynced them successfully (often many times). Every time you do a scrub you are reading all the sectors, and I do that every three months on my 15 TB, 16 TB, and 18 TB RAID-5 volumes with no problems. The math says that just won't work.
So it's clear that you can't just blindly apply the math. The error rate is spec'd at < 10e14, and the "less than" clearly is important. Drives are much more reliable than that spec suggests.
Yes, the drives must perform better than suggested by the specs. Otherwise the rebuild of a >12TB array is pretty much guaranteed to fail. Maybe the errors occur in batches rather than independently?
Once I am ready to rebuild what do I need to do? Is it a case of powering down, removing all partitions from the drives, plugging them in the right slots and powering back up? Does the NAS store any state in its flash?
There isn't much state in the flash (and no personal information or configuration data).
Unformatting the existing 3 TB drive and then powering up with the new drive mix in place will work. You can also use the boot menu to do a factory install (without unformatting the existing drive).
Though you might want to test the new drives (lifeguard or seatools) before you install. And it wouldn't hurt to run a long diag on the 3 TB drive while you are unformatting it.
hp532 wrote:
... Maybe the errors occur in batches rather than independently?
In my experience they do. Once I get one URE, the count usually starts to rapidly climb. Normally I'll replace a disk before reallocated sectors+pending sectors reaches ~50. Unfortunately OS6 uses much higher thresholds for email alerts (and this is one case where OS 4.x is better).
couple more thoughts;
- if you use the 3tb drive with the 8tb drives, make sure either all the drives are factory defaulted together, or you start with the 3tb first. --- you cannot add a smaller drive to an array with larger drives.
- StephenB even though you maintain multiple backups with your single redundancy, consider the time it will take to rebuild and restore your data if the single redundant main array failed in any manner. dual redundancy offers that extra protection against the, some would say, inevitable. You can of course potentially swap a backup device into a primary, but that still leaves you without one of your backup (and still potential lost data if the backups were not perfectly up to date).
NETGEAR Academy
Boost your skills with the Netgear Academy - Get trained, certified and stay ahead with the latest Netgear technology!
Join Us!
