We've got a lot of storage testing cooking at the PC Perspective offices, and while I usually hold off on publishing things until all testing is complete, I found myself merging two new products in a way that just begged for a photo and quick status update post:
This is a Drobo B810i on our test bench being loaded with 64TB of Helium-filled Western Digital Red 8TB goodness. I made it a point to evaluate this capability since Drobos have historically been limited to 16TB (or 32TB) maximum volume sizes. Drobo has been rolling out firmware updates enabling the new 64TB volume size in units with sufficient performance and bay count to support it (starting with the B1200i last year, and most recently with the 5N). This test was mainly to confirm the B810i's 64TB maximum volume size. The end result looks something like this:
With single drive redundancy (a minimum requirement for any Drobo array), the available capacity comes in at just under 50TB.
Dual redundancy mode drops available capacity down to just over 43TB. Not too shabby considering the Drobo can sustain two drive failures in this mode.
Drobo testing is still in progress and will take a bit more time, but I've completed the initial round on an individual 8TB WD Red and will be posting that review up shortly. Speaking of which, I'm off to get back to it!
I know nothing about this
I know nothing about this stuff. Can someone explain how you can have more than 32TB of storage with 64TB of Hard Drives and still have redundancy?
Look into RAID. Drobo’s
Look into RAID. Drobo’s BeyondRAID is a fancy RAID5, which is stripe with parity. You create parity bits to reconstruct data when you have drive failure. It’s not what you think it is.
If you apply a mathematical
If you apply a mathematical XOR across the same address across all drives (minus one), the result of that equation (parity) can be saved to that same address on the last drive. The same process is repeated in a rotating manner, placing the parity result on a different drive after the end of each 'stripe'. If a drive fails or is otherwise unavailable (array degraded), the same XOR can be applied to the remainder of the drives, and the result of the equation will be the data from the missing/failed disk.
The above is how RAID-5 (single drive failure protection) works. RAID-6 (dual drive failure protection) is similar in principle but uses more complex math.
These systems are more capacity efficient than simple mirrors (two of everything) to achieve redundancy against drive failures. The catch is that parity data must be calculated on-the-fly during writes (and during reads if the array is degraded). This overhead creates an additional bottleneck in the pipeline when it exists. Modern RAID processors can hardware accelerate these calculations, making the potential overhead less of an issue.
Are there any types of NAS
Are there any types of NAS Systems/RAIDS/Configurations that allows me to add as many drives as I want? I currently have many different drives, some new some old, with varying capacities. Is there a possibility I can create one expansive virtual disc/folder, similar to JBOD, while having some form of redundancy and easily add/remove drives by hot swapping, while letting the system handle the redistribution all on its own?
It’s obvious large scale enterprise data centres have these, but are there any dumbed down versions that’s easy to use for average techies?
My plan is to just buy or setup a simple NAS, start with three to four 4TB drives, then later add in 6TB and 8TB drives, while having them all running together seamlessly. The setup doesn’t have to be fast, as long as all the drives work together to form one single virtual drive/file, while maintaining some form of redundancy, then I’m happy.
You can do that, but not with
You can do that, but not with typical RAID or NAS. Must build your own box with drive pool configuration. Technically drive pool has no size limit (PB of data). Look at StableBit Drive Pool or Drive Bender. You can add/remove any drive you want and go with redundancy through the roof. Something that no RAID subsystem can do. 3x redundancy have some, 5x redundancy yes why not, 10x no problems!
That’s why NAS was never really option for me. Pooling drives is so much simpler and efficient (and much cheaper). Biggest strength of DP is that if drive fails or pool fails you don’t lose all data. Only failed drives. NAS box can fail very easily and then you’re screwed because of proprietary software. Keep that little detail in mind.
There are various ways of
There are various ways of doing this (Drobo units can handle mismatched drives with parity-based redundancy), but the software solutions that can work with varying drive sizes tend to lean towards mirroring over parity. If you can deal with the mirroring aspect, StableBit DrivePool will let you add your drives (with their existing data staying in place, provided they are NTFS formatted drives), and it can then merge all files into one large volume. It can then add redundancy by mirroring to other drives (at the file level).
Another thing I like about DrivePool is that as you add higher capacity drives, you can direct the removal of older / smaller / slower drives and the software will simply shift the files over (on the fly) and then dismount the drive.
I'm sure others can comment on other solutions.
Have to admit DrivePool is a
Have to admit DrivePool is a nice solution. Got it after your various mentions of it. Got Drive Pool, cloud component too. though still having performance issues that I need to properly configure.
Will admit their team is very helpful when I do have questions on the setup.
Have you tried to mix NAS/Drobo/etc with DrivePool as a singular solution? Tips and tricks would be welcomed 🙂
Many modern NAS these days
Many modern NAS these days will do this..
So the current offereings from Drobo, Thecus, Synology, QNAP, or Asustor should have no problem with adding capacity with redundancy with any mix of drives..
Synology and QNAP are the “big” names for home NAS systems, with Drobo and Thecus also big, but fewer features.
Asustor has the features but is the new kid on the block (at least in the USA).
Brands like WD, Seagate, Netgear etc probably won’t handle different drive sizes in parity/redundant mode without losing a lot of capacity.
Many modern NAS these days
Many modern NAS these days will do this..
So the current offereings from Drobo, Thecus, Synology, QNAP, or Asustor should have no problem with adding capacity with redundancy with any mix of drives..
Synology and QNAP are the “big” names for home NAS systems, with Drobo and Thecus also big, but fewer features.
Asustor has the features but is the new kid on the block (at least in the USA).
Brands like WD, Seagate, Netgear etc probably won’t handle different drive sizes in parity/redundant mode without losing a lot of capacity.
my understanding is it works
my understanding is it works roughly like this
Lets say you have 10 HDD, 9 for storage, 1 for backup
lets compare all the first sectors in all the HDD, and sum it.
The sum of Sector 1 is, for example, 227.
IF we loose one drive, the new sum is 200.
we can reconstruct the damaged drive by simply subtracting the first number from the second, so the first sector should have been 27.
do this for every sector, and you can lose any ONE drive, if you run into a bad sector or a multiple bad drives, you are hosed.
Yes, but add some high speed
Yes, but add some high speed tape backup in the background to backup the drives to the tape/s for extra protection should any unforeseen events may happen(multiple bad drives, those cheap Christmas lights igniting and toasting the house and all the drives. Especially for those necessary TAPE BACKUPS stored in the underground water/fire proof EMP shielded vault!
Really, RAID is for getting your Random Data Access/storage back up as quickly a possible if you need uninterrupted service, but you also need Tape backup, preferably tapes stored at a hardened alternative/different location should all hell break loose!
I prefer Synology units, but
I prefer Synology units, but this system looks hot…
64TB raw is also drool worthy,
We’ll have something right up
We'll have something right up your alley here very soon.
remc86007,
Not all redundancy
remc86007,
Not all redundancy is mirroring which may be what you’re thinking of when you ask that question. You also have, as an alternative to, or used in conjunction with mirroring, parity bits striped across the drives, such that it you lose a drive, or more than one in this case, the data is not lost, and the parity will be recalculated and striped again once the failed drives are replaced.
Just lookup RAID and the various levels such as 5, 6, etc. but know that these vendors sometimes utilize a proprietary method for redundancy, or use RAID-Z from ZFS which is a bit different.
So is that setup cheaper than
So is that setup cheaper than a Tesla? How much are those drives?
$300 for 8TB.
$300 for 8TB.
this vs your home grown nas
this vs your home grown nas please
This needs a FreeNAS build.
This needs a FreeNAS build.