This topic has been marked solved and closed to new posts due to inactivity. We hope you'll join the conversation by posting to an open topic or starting a new one.
I currently have 5 of my 8 drive bays all filled with identical HDDs. They are configured in "X-RAID". I have two questions:
1. What is the difference, if any, between X-RAID and RAID 5?
2. I currently get 4 x my drive size as usuable storage space. If I add a sixth HDD, will I get 5 x my drive size as usuable space, or will the NAS automatically convert to RAID 6?
1. What is the difference, if any, between X-RAID and RAID 5?
X-RAID is the auto-expandable RAID technology that is available only on ReadyNAS systems. It uses the standard RAID levels, RAID 1, 5 and 6.
With two disks, it will use RAID-1. With three disks, it will use RAID-5. With 4 or more disks, you can optionally select X-RAID with dual redundancy(RAID-6).
With X-RAID, you can add storage space without reformatting the drives or moving the data from another location. You may check this link for more information about X-RAID.
2. I currently get 4 x my drive size as usuable storage space. If I add a sixth HDD, will I get 5 x my drive size as usuable space, or will the NAS automatically convert to RAID 6?
If the volume is configured with X-RAID with single redundancy (RAID-5), it will not automatically convert to RAID-6 when you add the 6th drive. Changing to X-RAID with dual redundancy (RAID-6) would require you to destroy and recreate the volume.
1. What is the difference, if any, between X-RAID and RAID 5?
X-RAID is the auto-expandable RAID technology that is available only on ReadyNAS systems. It uses the standard RAID levels, RAID 1, 5 and 6.
With two disks, it will use RAID-1. With three disks, it will use RAID-5. With 4 or more disks, you can optionally select X-RAID with dual redundancy(RAID-6).
With X-RAID, you can add storage space without reformatting the drives or moving the data from another location. You may check this link for more information about X-RAID.
2. I currently get 4 x my drive size as usuable storage space. If I add a sixth HDD, will I get 5 x my drive size as usuable space, or will the NAS automatically convert to RAID 6?
If the volume is configured with X-RAID with single redundancy (RAID-5), it will not automatically convert to RAID-6 when you add the 6th drive. Changing to X-RAID with dual redundancy (RAID-6) would require you to destroy and recreate the volume.
Unless something has changed in recent OS versions, adding a 7th drive will convert to RAID6 unless you first switch out of XRAID.
Are you saying that the volume would change automatically to RAID 6 without data loss? Or would you be forced to create a new RAID 6 volume?
I've always wondered what the maximum number of HDDs are that could be configured in RAID 5. You must be saying it is 6? Clearly you couldn't build a RAID 5 array with 15 HDDs and only a single HDD used for redundancy.
Unless something has changed in recent OS versions, adding a 7th drive will convert to RAID6 unless you first switch out of XRAID.
Are you saying that the volume would change automatically to RAID 6 without data loss?
Yes (which also means there would be no increase in capacity). Note this does assume that the biggest three disks in the RAID-5 array are the same size.
I've always wondered what the maximum number of HDDs are that could be configured in RAID 5. You must be saying it is 6? Clearly you couldn't build a RAID 5 array with 15 HDDs and only a single HDD used for redundancy.
There is no maximum, and in fact you can build a RAID-5 volume with 15 HDDs. I'm not saying that's a good idea, but it is possible. Netgear simply decided (somewhat arbitrarily) to have XRAID transitition to dual redundancy when the 7th disk was added in a desktop NAS. My recollection is that the rackmount models switch to dual redundancy when the 6th disk is added.
One clarification - with RAID-5 the parity blocks are not stored on one disk. They are distributed evenly across all the disks. That improves write performance, and also balances the disk I/O evenly across all the disks.
One clarification - with RAID-5 the parity blocks are not stored on one disk. They are distributed evenly across all the disks. That improves write performance, and also balances the disk I/O evenly across all the disks.
Well I didn't know that fact. I also didn't know that in theory there is no maximum number of HDDs in a RAID 5 array. So with that understanding (just for fun) can the total usable space of a RAID 5 array truly always remain at total number of discs minus 1? It seems to me that if your number of discs got high enough, say 100, that the amount of distributed parity data must eventually equal more than a single disc size.
I don't know if there is a hard limit on raid-5, and if so it would probably vary depending on the particular implementation (ie linux may allow so many, while windows supported a different amount, plus there are hardware implementations (ie add in raid cards) which would have a hard physical number of drive it would support.
All that said, no one in their right mind would do 100+ disks in a raid 5 array, as the rebuild/sync time would be approaching infinite, and if you lost more than a single disk at the same time or during a rebuild/sync, you would loose all data across the whole set.
raid 6 doubles the redundancy so that if 2 disks fail you still have your data, but if a 3rd disk failed then again you would loose everything.
In the case of super large sets of disks, they would normally be in some kind of tiered approach, using multiple sets of disks in differing levels of raid, IE raid50 or raid60, where the parity stripes are constricted to specific sets of disks, resulting in increased redundancy (and accordingly less efficient total storage).
I think most would recommend a maximum of 24 disks in ANY array, but raid6 would be highly recommended for anything over 6 disks.
For example with sufficient disk slots, you could create different tiers with different performance and reliability, for example a 24 disk server:
- 4 disk raid 5 SSD perhaps for OS and apps
- 2 disk raid0 scratch disks for max performance temporary data
- 6 disk raid5 for high performance but with less redundancy (perhaps for stuff that gets modified and/or accessed frequently and/or by large number of clients)
- 12 disk raid 6 for regular performance with increased redundancy (perhaps for archive stuff that is stored and accessed less regularly)
obviously this is not designed for maximum storage, but for varied levels of performance spread across multiple disk sets and arrays.
finally, one should thoroughly understand that *NO* type of raid is a replacement for backups.
backups = multiple COPIES of your data, ideally on different physical devices, and even multiple locations.
consider not only disk/hardware/software failure, but also user error, virus/malware, malicious users, fire/flood/acts of nature, and physical theft.
Thanks TeknoJnky. My future goal is to back up my ReadyNAS in total as you describe. Currently I have specific folders backed up on various external hard drives located in my house. It is a clunky, manual system. I don't truly trust cloud storage or mainly don't want to spend the money. But as my NAS grows, it almost becomes necessary to build another NAS just to use as the backup. That gets expensive. I currently have about 11 TB of data on my NAS. I guess I could compare the cost of perpetual cloud backup to the cost of building the 2nd NAS. 🙂
So with that understanding (just for fun) can the total usable space of a RAID 5 array truly always remain at total number of discs minus 1?
Yes, it does.
RAID is organized into "stripes" that cut across all the disks.
If you had 100 disks with RAID-5, then each stripe has 99 data blocks and 1 parity block. For the first stripe, the ReadyNAS puts the parity block on the disk 100. The second stripe's parity block is on disk 99 ... Stripe 100 has the parity block on disk 1 - and then the pattern repeats.
When you resync after a failed disk, you recreate the missing block in each stripe from the other 99. You do that by doing a XOR (exclusive or) of the remaining 99 blocks in that stripe. If another disk fails during the resync, you can't do that computation - so the resync fails and the data is lost.
On the max number of disks - there likely are implementation limits on the number of disks in the array in mdadm. The SNIA Common RAID Disk Data Format Specification supports up to 4096 physical disks in a virtual disk, which sounds over-the-top to me.
Currently I have specific folders backed up on various external hard drives located in my house. It is a clunky, manual system. I don't truly trust cloud storage or mainly don't want to spend the money. But as my NAS grows, it almost becomes necessary to build another NAS just to use as the backup. That gets expensive. I currently have about 11 TB of data on my NAS. I guess I could compare the cost of perpetual cloud backup to the cost of building the 2nd NAS. 🙂
We have similar amounts of data.
I use Crashplan Pro for disaster recovery - protecting against theft, fire, flood, ransomware, etc. The cost is $10 per month for a single computer (e.g. the NAS) - which I think is reasonable. But if the restore speed matches the backup speed, then it would take several months to restore all the data. So there is still a need for local backup (even if you choose to fully trust the cloud provider).
I do use other NAS for local backup - although I agree it is more expensive than external drives. If you were starting from scratch, an RN212 + two 8 TB drives (jbod) would cost about $600, while two 8 TB external disks would cost about $300. So the incremental cost of using a second NAS could be as low as $300.
A "in between" approach is to get a two or four bay SATA enclosure for backup, and connect that to a PC. You can set up scripts (or use free tools) to wake up the PC on schedule and do incremental backups of the NAS.
