I got into it today on ZDnet with one of the other bloggers, George Ou, who published Why dumb-downed no-RAID storage is bad for consumers. As I believe that RAID is an idea whose time is coming to a close, I responded with Why home RAID won’t fly.
So far, ZDnet readers seem more persuaded by George
I’m in my trailer, sulking. How could they?
The exchange has sharpened my thinking, as George and some other folks came back with some good comments, and a couple of the more perceptive – obviously – folks came to my defense.
While I like the Drobo storage robot concept and Geoff Barrall personally, I’ll be very interested to see what kind of market they develop. Which is marketing-speak for “I’m dubious.”
Why?
The secular trend in computers is that technologies scale up from consumers – not scale down from the enterprise. But so what? The real question is why.
Because consumer stuff is cheap and enterprise stuff is expensive. Because one is high volume and the other low volume. Because volume enables low-cost experimentation and improvement. Because building cheap stuff usually forces people to focus on what really works for customers who won’t open a manual.
Home RAID? I don’t think so
Why not? Let me count the ways:
- Complexity: RAID fails ugly. Pick the wrong drive to pull or copy and your protected data is no more. And due to the redundancy, RAID systems have failures much more often than a single disk does.
- Completeness: while RAID solves some problems, it isn’t a substitute for a backup. Getting customers to understand that is hard. Not all the ZDnet readers get it.
- Cost: HW RAID means a controller, a chassis. A lot of money before you buy the first disk. SW RAID is cheaper – with Intel’s ICH8 chip almost free – and consumers still need to understand why they are buying a second drive and not getting more capacity.
The vast fetid swamp of consumer ignorance
In my small town I often help people with computer problems. Often these are small business people who’ve been using computers for years. What I’ve found is that these people don’t have the faintest idea how their computer works or how the components work together. To most people computers are magick.
Case in point: a professional photographer lives across the street. Two Macs, scanner, several photo quality printers, a couple of fancy digital SLRs. One Mac does color correction. The other is her main machine. Photoshop and a bunch of other image processing software that she knows how to use. Pretty sharp lady.
And she doesn’t know the difference between disk and RAM. It is all “memory” to her. She never added RAM to the skimpy amount Apple provided, so her disks are thrashed all day. She’d let the disks fill up, not realizing that she needs at least 10% free space just for the OS to use. A few hundred megabytes sounds like a lot to her.
This is the person you are going to sell RAID to? She’s your target market, with hundreds of gigabytes of valuable digital assets to protect. How would you start the conversation?
She does understand the value and process of making copies, which she would still need to do even if she bought your RAID gizmo. So how do you explain your value-add?
The StorageMojo take
Home RAID for the masses is an uphill battle. Backup is the battle the industry can win. What kind is the issue. Across the net to Mozy, Carbonite or some more fully featured option? Local backup to a DAS hard drive or to a simple USB-attached NAS drive? Those “one-touch” Maxtor drives?
Comments welcome, of course. Leaving for Boston today. If you’re in the neighborhood this weekend, send me an email and we’ll do coffee. I’ll be staying at Copley Square. Moderation may be a little slower than usual, but moderate I will.
This is an interesting discussion and thanks for the links to multiple papers. I am in a SOHO environment and am currently storing about 4 TB of data.
I see several advantages of RAID appliances:
(1) They allow larger volumes than a single drive
(2) They offer a degree of protection against single-drive failures
(3) The rebuild of a replaced drive is quite transparent to the user and relatively quick.
I use two main RAID arrays:
(a) is 5 x 1000 GB drives
(b) is 4 x 1500 GB drives
These are backed up in another building on:
(c) is 5 x 1000 GB drives
(d) is 4 x 1500 GB drives
There is a fifth array in my building for non-critical support files:
(e) 4 x 500 GB drives
All are running RAID 5 or a Netgear derivative.
Every night, the main RAIDs are compared on a share by share basis and the remote shares are updated, but deletes are not propagated.
About once a quarter, the main RAID is compared to a set of 2.5-inch USB HDDs in a steel ammunition case that is kept well off-site and appropriate updates are made. Depending on the share, some deletes are propagated. These drives are mostly 1000 GB, although there is a pair of 500 GB in there. Most are USB 3.0, the balance USB 2.0, and it takes a day to do the updates.
So, I have triple redundancy for most data. Two of the redundant sets are RAID-5 and one is stand-alone HDDs.
All the comparisons and updates are done through various ViceVersa Pro scripts–they have to be as some shares (images and audio) are larger than 1000 GB.
Here is an index to my blog posts on the subject which explains in more detail.
http://richardhess.com/notes/category/computer-data/data-storage/
Regards,
Richard
Hi all
First post here so here goes:-
Basically i think its all about training/teaching everyday uses the difference in the technologies and terminologies, i work in a IT Support Desk environment for business user and most of the calls we get are from normal every day business people who have been give a laptop/mobile/table…. but don’t know how to use it.
Do you think they read the manual that came with it? no of course not they just assume they know.
Same is true with home technology for RAID, DAS and NAS, I have been recently helping out my close friends and family members by training them on using their tech.
I have been creating RAIDS on there NAS’s which then Backs up to a 50 GB Drop box, ok i know its not as good as Mozy but its free and they can access from mobiles and the web.
Once they know how to use the Tech i do not get bothered as much…. which is very very good.
I have had this experience myself both in the office and at home. RAID is not useful for home purposes.
Recovery is possible; but only with professional datarecovery solutions. If your raid goes down; just bite the bullet and bring it in.
My small business (4 employees) doesn’t have an IT dept. I’m it. When our server “hiccuped” with a storm, our raid went down and I was crazy agitated.
Our server grade tape drive couldn’t recover using the server grade backup software because we never had the IT professional to setup, maintain and test it properly.
Our saving grace was a wonderful data recovery (100% complete) by cbldatarecovery and a “file copy” backup of the drive I had happened to make 2 weeks prior. Those were the only 2 backups that worked.
RAID IS ugly. Necessary in the office server but when it goes down; you need to find an expensive professional with RAID experience to guarantee a recovery before you screw it up yourselves.
In the home environment, I now do the google method. 3 copies of the same data at any time. I try to put all important data on 1 partition and weekly image that partition to a second drive. Every 3-6 months; I image out to an external usb drive and move it to the office/different location.
HOPEFULLY that should work.
I don’t know why every comment treats RAID* as though it universally treats data the same way.
Depending on the RAID type, it complements BU very well, but home users *could* get away with RAID only solutions with minimal fuss.
The main problem I’ve found using RAID is how controllers implement RAID differently, and depending on the FileSystem, power failures become an issue.
Solution for RAID-only? Use a PSU battery & know what RAID trade-offs there are for RAID types, and know what to expect out of the FS.
Most people I know (self included) use RAID for the speed benefit; that redundancy is a bonus.
For instance, RAID-0 tremendously increases data throughput speed, but is dumb to employ without a backup; If a drive corrupts, bye-bye data.
RAID-1 is essentially a BU (redundancy) without explicitly backing up, but doesn’t have the speed advantage; if a RAID-1 drive fails, put in a new one and let the repair copy from the BU, and you’re back on track.
RAID-01 combines both: speed with the ‘speed’ drives backed up.
RAID-5 combines both, but emphasises rebuild ability to reconstruct the array without requiring as many disks as RAID-01, and you can stay ‘live’ with your system.
It makes sense to demand a BU for JBOD and RAID-0, but for the others, it’s not necessary at all (unless you have a crappy RAID controller).
The hype I read here about ‘RAID is on the way out’ is simply FUD.
Know what tech you use vs what you need, and as another poster said, *learn* your equipment – especially is your income depends on your equipment!
The hype I read here about ‘RAID is on the way out’ or ‘RAID is ugly’ is complete FUD.
I also don’t know why every comment treats RAID* as though RAID universally treats data the same way.
Depending on the RAID type, it complements BU very well & can *be* the BU. Home users *could* get away with RAID only solutions with minimal fuss, if they bother to ‘read the manual’ and not just *assume* how their foreign equipment works (blog writers included).
The main problem I’ve found using RAID is how controllers implement RAID differently; Depending on the FileSystem, & controller, power failures become an issue.
I won’t advise someone Not to use a dedicated BU, but I personally don’t find it necessary (12+ years with various personal RAID configs; currently use RAID-5 for I/O intensive project files & document library).
Looking for RAID-only solution? Use a PSU battery & know what RAID trade-offs there are for RAID types, and know what to expect out of the FS, and use hdd disks designed for RAID/NAS. Most MoBo RAID controllers are crap.
Most people I know (self included) use RAID for the speed benefit; that redundancy is a bonus.
For instance, RAID-0 tremendously increases data throughput speed, but is dumb to employ without a backup; If a drive corrupts, bye-bye data.
RAID-1 is essentially a BU (redundancy) without explicitly backing up, but doesn’t have the speed advantage; if a RAID-1 drive fails, put in a new one and let the repair copy from the BU, and you’re back on track.
RAID-01 combines both: speed with the ‘speed’ drives backed up.
RAID-5 combines both, but emphasises rebuild ability to reconstruct the array without requiring as many disks as RAID-01, and you can stay ‘live’ with your system.
It makes sense to demand a BU for JBOD and RAID-0, but for the others, it’s not necessary at all (unless you have a crappy RAID controller).
You can find decent (and much above decent), dedicated RAID controllers that won’t break the bank. I also have an external e-sata/usb3 RAID enclosure that’s been really good to me for the past ~3 years (no bad disk yet), and I’ve travelled with it, and use it as a ‘BU repository’ for several of my machines, while enjoying the RAID performance from it.
Know what tech you use vs what you need, and as another poster said, *learn* your equipment – especially is your income depends on your equipment!
I think the other story is that the industry has failed to produce a backup solution that works for consumers.
My first computer was a TRS-80 color computer that used tape to store programs an files and I can say that my first test of tape was a disaster. Even though I had the manufacturer recommended tape deck and cleaning kits and everything, it was a crap shoot to save anything to tape and get it back.
I bought a floppy disk as soon as I could and never looked back…
Until I got kicked out of the computer center at my old school and they put my old files on a Suntape that I couldn’t read. And then I had a QIC tape deck for my computer that would have taken 10 tapes to back up my drive and still was unreliably. Then there was the place that I worked at which had a tape robot from IBM and it would take 15 hours to restore just one file (if a source code or configuration file was lost I always wound up re-creating it.) If I did try to restore a whole machine I’d inevitably find at least one file that was lost or corrupted.
Now there is network backup, and carbonite used to advertise heavily on the radio, but unless you have gigabit fiber, your network connection is not fast enough to back up your machine in a reliable way.
Just remember that our technology is not bullet proof and it will fail its only a matter of time. Never rely on it so make sure you have up to date adequate backup of your data.
I personally believe in RAID. I’ve been using it since 2001 and never lost a single bit of data since. This was for domestic use, and for my small business as an artist working in Photoshop to create comics.
The problem with any sort of backup is it’s never up to date. Even with daily backups, you lose up to a day of data when a drive fails. Not so with RAID.
The sense of happiness I had the first time a drive suddenly failed and all I had to do was wander down to the shop, buy another drive, pop it in and rebuild the array, compared to the “OMG” feeling of an un-RAIDed failure is all I need to convince me to keep using it.
I backup as well, because RAID can’t protect against some failure modes. But it hugely reduces the probability of losing work you can never recreate (created works like art, music, writing, etc).
One suggestion: when you do lose a drive, copy all the data off the RAID before starting the rebuild. Better safe than sorry.
Robin