External RAID enclosure

Community Forums/General Help/External RAID enclosure

GfK(Posted 2013) [#1]
I'm looking into getting an external RAID-1 enclosure for backups, as I figure I've diced with disaster for long enough. My current method of manually copying from dev PC>>USB drive>>external HD>>>DVD is fine, but it's all a bit of a pain.

So, RAID-1. Two drives, treated by the OS as a single drive, and the same data synced to both.

Some questions:

1. Do the drives have to be identical brand/size?
2. That being the case, what if one drive dies and I can't find a replacement? Am I as fubared as if I just hadn't bothered?
3. Any good brands of enclosure, or ones to avoid?

Ta.


D4NM4N(Posted 2013) [#2]
Check out the netgear NAS duo. Had one for a few years now. It has a simple but quite powerful web based interface and a couple of raid options, and some good backup tools.

It is not just a raid though it is actually a debian based mini-server (It would not be as fast as a dedicated raid with firewire cable, so may not be what you want. You can also install an SSH access client on it and actually access the nux undersystem and install things like git and apache too on it if you want)

They cost about 180-200 odd quid with 1x1TB disk installed (so add 80 quid for a second hdd)




xlsior(Posted 2013) [#3]
1. Do the drives have to be identical brand/size?
2. That being the case, what if one drive dies and I can't find a replacement? Am I as fubared as if I just hadn't bothered?
3. Any good brands of enclosure, or ones to avoid?



1) you typically get optimal speed if the drives are the same make/model, since different drives will have slightly different specs. If you use different drives, then the slowest drive will set the pace. Although if there are drastic differences in drive specs like RPM, platter counts, or data density, the drives could be completely 'out of sync' and worst case scenario you may see half the throughput that you could have gotten with identical drives.

another thing to keep in mind with using different drives in general, is that not all drives sizes are *actually* equal.
For example, my 3TB hard drive actually contains 3,000,579,911,680 bytes.
Fine, whatever. But if you buy a *different* brand/model 3TB hard drive, they may have a slightly different setup, thanks to different number of platters, subtle changes in data density, more or less space set aside for error correction, etc.
But hypothetically, the second (different) 3TB drive may contain 3,000,000,000,000 bytes.

when you *start* that is no problem (the enclosure will typically figure out the largest amount of space available on both drives, and limit tot that: e.g. the RAID volume would be 3,000,000,000,000 bytes.
But if you had started with two of the first drives, you'd have a 3,000,579,911,680 byte volume, and buying a brand new replacement 3TB drive down the road could still present a problem when there is no room on that 3,000,000,000,000 byte new drive to mirror the 3,000,579,911,680 bytes of existing data: won't work.

(Been hit by that several times in the past myself, where the replacement drive is a few MB short of working)

So, pro-tip: If you have a RAID enclosure where you have ANY kind of control about the size of the RAID stripe, I'd strongly recommend not allocating a full 100% but build in a little bit of headroom to safeguard for a future disparate drive, e.g. leave 1% unallocated to make sure that the actual array won't be larger than the replacement drive might be.

Also, there are a LOT of cheaply made, very poorly supported unreliable devices out there. READ REVIEWS!
Pretty much, anything cheap is pretty much a piece of junk. reliable hardware raid is not that easy, and you really don't want something from a company that would continue to cut corners to save another buck when your data is at stake.
Keep in mind that RAID is a combination of many different things: you need reliable software, but also a good power supply / adapter, decent backboard, good quality capacitors on the motherboard, etc.

Personally I'd stay away from the netgear, dlink, and other router companies RAID devices because it really isn't their core business, and overall they appear not very well made.

I'd suggest looking at devices made by companies like qnap, that have been specializing in this market and make both high-end enterprise devices as well as home/small office models. At least they'll take the problems involved a bit more seriously since shoddy home units could damage their enterprise reputation as well.

Yes, they aren't cheap. Neither is losing your data, I'm sure.

finally, keep in mind: RAID is not a backup. RAID is a convenience that will help your survive certain hardware failures, but it's not the end-all of data rescue: you still should have additional (offline) backups to warrant against lightning strikes, virus attack, user error (oops! didn't mean to delete THAT folder!), water damage, etc. etc. etc.


xlsior(Posted 2013) [#4]
Oh, and another word of warning: If you're looking around for harddrives, you'll find that there are a number of them specifically labeled 'RAID' or 'enterprise', which are typically quite a bit more expensive than the 'normal' non-raid/non-enterprise versions.

Part of the issue is that the enterprise drives are typically engineered to a higher standard -- after all, they are made to be operated 24/7/365, with zero downtime, yet stay reliable. Most 'consumer' drives are officially rated for 8-10 hours of operation a day, and may have a lesser lifetime when used 24/7. the enterprise drives
There may (or may not) be physical differences (bearings, platter material, coatings, quality of internal components, etc.) but more importantly: There are differences in the drive firmware that can make or break a drive in a RAID environment.

Many of the consumer drives these days are marketed as 'green'. Sure, Fine, whatever. But part of what makes them 'green' is lower power use. Great, everyone wants to save battery life in their laptop. But how does it accomplish that? By having the drive controller automagically spin down after some time of non-activity, without waiting for the OS to order it do so. However, since the drive did it on its own, the RAID controller didn't know about it either. Then when you really want to read/write again, there is a delay,w aiting while the drives spins back up. Hey, that took a really long time. Perhaps the drive is bad? The drive must be bad. Error, error! You may have a predicted drive failure. Hey, it keeps happening! That drive must be bad. Let's drop it out of the array and wait for a replacement!
(google around, you'll find lots of people with horror stories about their 'green' drives not working reliably or for extended periods in their enclosures)

another issue: Even if the consumer drive isn't a 'green' model, some of them could go in a diagnostic 'reboot' every so many runtime hours. Perhaps after running uninterrupted for a month or so, your consumer drive may do a quick 30-second diagnostic on itself or something. No problem in a single drive setup (your machine may stall for a bit, then continue on merrily on its way). but in a RAID setup, such a stall could cause it to think the drive has failed (or is about to fail), and drop it out of the RAID array breaking your redundancy.
Another drive may do it after a week. Some may never do it. But you won't really know until it's too late.

Bottom line: for *reliable* hardware RAID, you really shouldn't use non-enterprise harddrives.