RAID - Need advice understanding what would be best for me.

[tingies213]

Roughly two months ago I built a new computer with three 1.5TB drives and installed Windows Home Server to use it as a media server. I was very happy with how things were going until two weeks ago I discovered that one of the data drives had a lot of bad sectors. I managed to salvage most of the media off of it before I RMA’d it but I lost a good chunk. The other data drive was full so when I transferred all the salvaged data it was written on the system drive.

Yesterday I started transferring 250GB worth of mp3 files from my main computer to the server and left to attend to something else. 20 minutes after I started transferring I went to watch something on my HTPC in the living room which pulls media from the server but there was no media to be found. I went and reset the server without bothering much about it and went back to the living room and waited for it to boot with no avail. I went back and turned the screen on to see that one of the drives was having problems in BIOS. I managed to boot it up eventually and ran chkdsk with bad sector detection on. It took several hours to finish and I wasn’t there to see the result. When I logged on the WHS console I saw that the system drive had 392GB worth of system data. I don’t know what that data is supposed to be but it shouldn’t be there. I started transferring files and the transfer rate was in the Kb/s for several minutes. I would stop and start again to see if something would change. Still nothing, transfer rate was very slow. So I just left it there for a while to see what would happen. After several minutes it reverted back to normal transfer rates (more than 20Mb/s). I know there’s something wrong with the drive but since I have all the data backed up somewhere I left it there for now and started thinking that this might not be a good solution for me.

I’ve started reading about RAID and all of its different configurations. I’ve never had any experience with RAID so my knowledge is very very limited but from what I’ve read I think that building a RAID array with redundancy would be the best solution for me since I value my media. I don’t care one bit about performance; all I care about is the safety of my media. I see 3 possible solutions, RAID 5, 6 or maybe even 1+0, please correct me if I’m wrong (I’ve excluded 0 because although I think it’s the safest it also eats up too much space for my liking). I can’t decide which would be better suited for my needs. Is it harder to build one array compared to a different one? Is it more difficult to build an array with more hard drives? Does hard drive space matter? Can I add drives as a I go or do I need to decide how much space I will need and have all the drives ready? That’s one of the features I absolutely love about WHS, adding drives as I go. Will RAID work in WHS? Would I be better off with a different OS? How much space will be used for redundancy? What about hardware? What kind of hardware will I need depending on what I choose?
I know I have a lot of questions. This is probably a long term project and I need to be sure I do it right. Once I set it up I don’t want to bother with it much like I’m doing now with my current setup. Most importantly I want my data to be as safe as possible without sacrificing too much space.

Any advice would be greatly appreciated.

 

[alzan]

My knowledge of RAID is limited to levels 1 and 5, plus some general tidbits picked up over the years; now that I've made that disclaimer, on with the advice.

A good place for general RAID info is Wikipedia's article.

It's not necessary to have all the drives at once, just remember that when you replace one in the future it needs to be the same or larger size than whats in the array.

I don't know enough about WHS to answer if RAID works with it. My gut reaction is yes.

How much space is used for redundancy depends on the RAID level: RAID 1, minimum of two disks, one disk will be used for the redundancy. RAID 5, minimum of three disks, one disk is used for redundancy (actually one disk worth of space is used, since the parity information is spread across the disks in the array.)

Depending on your case and controller you can also set up an extra disk as a hot spare to take the place of a disk that dies.

If you're serious about using it, be prepared to spend money for good hardware. A good controller, not just one built into the motherboard, and good drives. Good controllers would be LSI or Adaptec, they will also support more different RAID levels than an onboard chip; drives should be enterprise-class, like the Western Digital RE3 & RE4 series.

Stay away from RAID 0 unless you keep backups and don't mind the downtime while you restore. It is the fastest in terms of write/read; but has no redundancy. RAID 1+0 or 0+1 would be alternatives, although I would think more expensive in terms of disks.

RAID 1 would give some read performance benefit & redundancy but the least amount of space in the array.

RAID 5 would give you some performance benefit and redundancy, it will also, depending on drive size, give you more usable space in the array.

Hope this info helps; or at least gives you a starting point.

alzan

 

[RebateMonger]

Most importantly I want my data to be as safe as possible without sacrificing too much space.

And that's what you will have to decide. Safety and space efficiency

The safest is to use backups. That takes 1:1 disk space.

The next safest is to use RAID 1 or WHS' built-in Folder Duplication. Both use use 1:1 disk space. WHS' Folder Duplication is the easiest and allows you to easily add additional disks to WHS.

RAID 5 is the most common "space efficient" form of redundancy. But if something goes wrong (which it definitely can), you lose everything in the entire RAID 5 array.

You can use RAID underneath WHS, but it does add complication and you can't just toss in an additional storage disk.

The "best" answer depends on what's most important to you and how much your data is worth to you.

 

[Paperdoc]

Forget about RAID0. It is the one in which your data is split over 2 drives and BOTH MUST function for anything to work. One bad drive = NO data!

RAID1 does constant mirroring - there are always two compete copies of everything on separate drives. It gives you only half of the total drive space to use, though. It is great if you MUST be able to keep running when a drive fails, BUT it is NOT a substitute for a backup system.

RAID5 requires at least 3 drives, and the classic form is five. It splits data into four chunks and then calculates a checksum set before writing all five chunks to disk. The data are written to different disks. If you have five disks and one of them fails, there is enough information on the remaining four to completely rebuild ALL for your data once the faulty drive is replaced. I'm not so sure about that recovery detail for less than five disks, though. The recovery does take a lot of time, though, and the extra work makes the whole disk system a little slower than a single drive. You get to use about 80% of the total disk space in the drives you use. However, if TWO drives fail at the same time (and I have seen this!) you do lose everything!

RAID6 takes more drives that RAID5 to give you the same usable space, but it can restore itself even if TWO drives fail.

NONE of these options is as good as a proper backup system, especially for what you want. You want to be sure you never lose your data. Well, that is best done with regular backups. AND you must make the backups and then VERIFY that they are valid copies you can use to restore from. Then you REALLY need to disconnect the backup unit from everything and store the media offsite. So look closely at a good backup system. It will probably do a better job for less money than a really secure RAID5 of RAID6 system.

 

[JackMDS]

The problem of Enthusiast is lack of understanding the term "Time to Let Go"".

If there were 6 reasons in the past to use RAID for General Purpose Workstation and Backup. Four out of these 6 are Water under the Bridge.

Or in other words, with current status of Space/Speed/Price in most cases there is No reason to employ RAID any more.

 

[Nintendesert]

The problem of Enthusiast is lack of understanding the term "Time to Let Go"".

If there were 6 reasons in the past to use RAID for General Purpose Workstation and Backup. Four out of these 6 are Water under the Bridge.

Or in other words, with current status of Space/Speed/Price in most cases there is No reason to employ RAID any more.

Considering how cheap large HDDs are, it would probably be less of a headache to simply purchase a few and back up to them and store the drives in a safe place.

That way you're also protecting your data from environmental hazards, which a RAID won't do.

 

[elconejito]

If you're going to continue to use WHS, your best bet is folder duplication as RebateMonger mentioned. It takes any folder you specify and duplicates that folder on another disk. It's like RAID1, but at a folder-level instead of a drive-level.

Then just ensure that you have enough space. So if that folder has 250GB worth of data, make sure that the other drive has 250GB of space also.

Doing it this way lets you continue to use WHS which is great for sharing media across your network. And it let's you just add many more drives along the way, in varying sizes. WHS will figure out when, where, and how to duplicate the data across whatever drives you have.

Otherwise, you could RAID the drives (RAID1 or RAID5) which would add another layer of complexity (you'd RAID outside of WHS). RAID1 you need two identical (in size at least) drives to "mirror" each other, or RAID5 you need 3 drives. Either one will survive the loss of any single drive.