bpulv wrote:
This is new information. I have identified it as RAID-2 based on my apparent misinterpretation of definitions of the different RAID modes I have seen on the Internet.
As far as what I have said about the Drobo function, your explanation is not the impression I got from the Drobo website when I purchased it. It was my understanding that each drive was a mirror image of the other drives. I know performance is degraded with I swap drives, but that has never been a concern because I run the Drobo over night and the next day it is done. One time I had a hard drive fail. The red light flashed on the Drobo to warn me. When I replaced the drive with a new one, the Drobo rebuilt the new drive and then went back to normal operation as advertised.
This is new information. I have identified it as R... (
show quote)
Just don’t want to see you (or anyone else for that matter) get burned when you really need your backup to function.
First, RAID is a good thing - it adds redundancy or speed or both compared to a single drive. Here are the commonly used RAID levels.
RAID 0 - simple striping of data across multiple drives. You gain speed, but if any drive in the stripe fails, you lose everything - generally a bad idea.
RAID 1 - simple mirroring. Adds redundancy, but uses 50% of the space for it. Needs 2 drives minimum.
RAID 0+1 (or 1+0) - striping and mirroring. The favorite of DBAs. Roughly doubles the speed and adds redundancy, but still uses 50% of the storage for redundancy. 4 drives minimum and can survive a double drive failure DEPENDING on which two drives fail.
RAID 3 - byte striping with a dedicated parity drive. The best choice for large files such as imaging, but rarely available except for film/video applications.
RAID 4 - block striping with a dedicated parity drive. A good compromise between small file/large file performance. Mostly used by NetApp (Network Appliance) with their WAFL file system.
RAID 5 -block striping with rotating parity. The most popular current RAID level - allows multiple users to access different files simultaneously. 3 drives minimum. Can tolerate a single drive failure. 50% performance hit while in “degraded mode” while rebuilding data from a failed drive onto a spare drive,
RAID 6 - not a defined RAID level, but generally used to mean a RAID 5 with dual parity so you can survive a double drive failure.
There are other “made up by marketing” RAID levels like RAID 50 by small (non-commercial) RAID manufacturers of which there are dozens (like Drobo).
Now back to your specific case. Drobo allows simple mirroring (RAID 1) with 2 drives, RAID 5 with 3 or more, and “RAID 6” with 4 or more drives, which they combine into their “BeyoundRAID” virtualization. When you remove a drive, what’s happening is that you’re failing a drive (simulating a failed drive), and when you replace the drive with your rotating “spare”, the RAID goes into degraded mode and begins rebuilding the data onto the “new” drive. Depending on the nature of the rebuild (can it use the existing data on the drive or must it start from scratch), this can take from minutes to days with large drives. If another drive fails during the rebuild, there will be data loss.
What you should do is keep the 5th drive as a spare (forget the lock box trips), use the existing drives as you have been and replace a drive with the spare only if a drive fails. Also keep an off-site copy for DR (disaster recovery) as you have been, but do one of the following. (A) If you have good internet access, backup to a major cloud provider. The initial “seeding” of the cloud will take awhile (perhaps days), but afterwards you’ll upload only changes and new files (B) use a large off site drive as you have been. The drive needs to be large enough to hold the ENTIRE contents of the RAiD. Just mount it as an individual drive (not in the RAID) and either copy the data or use a backup or mirroring ap to regularly replicate the contents of the RAID onto this off site spare. Do this regularly (C) if not too large, do an initial and then incremental backups to MDisks (they are available up to 100GB each) which you move off-site (D) mirror your data to another system either on-site (which won’t protect you from a fire) or to a relative’s system at a different location (essentially the same as cloud storage, but much less reliable). Other than tape, which is impractical for individual users, these are the only methods I know that are reliable.
I hope this helps with your backup/DR strategy. You have a RAID system, which is certainly more reliable than a single disk. Now you just need to implement a good off-site DR/backup plan that you test to make sure it works BEFORE you need it. Personally, I mirror data to another on-site system, keep a copy of data in the cloud (Amazon S3) and archive important data on MDisks in a fire proof safe on-site.
Cheers.