RAID vs Scheduled Back-ups w/ External Hard Drive

[Greg]

I'm starting research into building a media server. In fact this would be my first computer build of any kind............zero experience. Any comments would be appreciated.

What would the PROS and CONS be for a RAID vs an extrenal hard drive with scheduled (nightly) back-ups.

On the CONS side for the RAID controller:

- increased read and write times due to accessing multiple drives
- complexity
- cost

I don't know enough to address the PROS over the external hard drive approach.

Oh, just thought of one: RAID handled your back-up in real time. With the external drive approach the data between scheduled back-ups is at risk.

Hopefully, you can help me understand.

Thanks,
Greg

[GKusnick]

Apples and oranges. RAID is not backup; RAID is a way to minimize downtime when a single drive fails. Real backup involves keeping copies of your important data in separate physical locations so that a catastrophic failure (fire, flood, earthquake, lightning strike, etc) doesn't wipe you out. RAID offers very little protection against that kind of disaster. On the plus side, it does (usually) give you the option of swapping out a defective drive without powering your system down, so that clients never notice the failure.

So the first step is deciding what data you're trying to protect against what sort of threat. Personally, I don't bother backing up recorded TV content; I'm OK with losing that stuff if a disk dies. I do back up my entire SageTV installation directory (among other things) so that the work I put into configuring my system isn't lost. I also keep copies of purchased video and music downloads at an offsite location so that investment isn't lost. For that sort of thing, an Internet backup provider might be a better option than an external drive (but it doesn't hurt to have a multi-tier backup strategy).

[Greg]

Originally Posted by GKusnick

On the plus side, it does (usually) give you the option of swapping out a defective drive without powering your system down, so that clients never notice the failure.

Thanks, Greg.

"Hot swapping" a drive to keep 100% up time is not all that important to me. We're getting along just fine right now without recording anything, but who knows after I spoil 'em with SageTV..........I'll be the on-call IT guy.

Thanks again,
Greg

[jsonnabend]

Greg is entirely correct, but he overstates the issue a bit in practical terms.

RAID is not backup in the sense that it does not protect against accidental deletion of files and catastrophic failure (like the house burning down).

Raid does act as backup against single drive failure (or double drive failure in RAID 6). That is, if a single drive fails and it is unraided, you lose your data. In RAID 1, 5 or 6, you don't lose any data. That's essentially the same function served by a backup.

With that in mind, especially with the low cost of network raid enclosures/external raided drives, it makes sense to setup raid and perform routine (real) backups.

My $.02.

- Jeff

[Taddeusz]

Originally Posted by jsonnabend

Raid does act as backup against single drive failure (or double drive failure in RAID 6). That is, if a single drive fails and it is unraided, you lose your data. In RAID 1, 5 or 6, you don't lose any data. That's essentially the same function served by a backup.

Not true. If you have a lot of static files, as is the case of many of our SageTV setups, bit rot can creep into the parity data of a RAID 5 array or into the drives of a RAID 1 array. Your RAID rebuild will fail in these cases.

Or, depending on the size of your array, a second drive failure during rebuild will take down the whole array. After a single drive failure the likelihood of a second drive failure goes up quite a bit.

Again, RAID is not a backup.

[GKusnick]

Originally Posted by jsonnabend

Raid does act as backup against single drive failure (or double drive failure in RAID 6).

I still wouldn't consider that a backup since it doesn't save you from needing a real backup solution. What it does save you is the need to restore from backup in the common case of single drive failure. RAID's data redundancy is how it achieves its function of minimizing downtime, not a true backup function.

[stanger89]

Originally Posted by Greg

I'm starting research into building a media server. In fact this would be my first computer build of any kind............zero experience. Any comments would be appreciated.

What would the PROS and CONS be for a RAID vs an extrenal hard drive with scheduled (nightly) back-ups.

If I limit my answer to the confines of pros and cons assuming the intent is to protect against having to re-rip (since the backup is the original discs):

On the CONS side for the RAID controller:

- increased read and write times due to accessing multiple drives

Good hardware RAID is actually faster than single drives, potentially much faster. A good RAID array can hit several hundred MB/sec read or write speeds.

- complexity

That is debateable IMO, yes you need a RAID card, but that's really little more than a hard drive controller. You need some more knowledge to set it up, but once that's done, it's actually simpler IMO because it looks like just one device instead of many.

- cost

This too is not straightforward. The reason is RAID5 and RAID6 (which meet redundancy needs easily for home use) are more efficient than duplication, becoming up to almost twice as efficient as the number of drives increases.

If you start getting into lots of data with lots of drives, the cost of all the extra drives for duplication can easily become greater than the cost of even a good hardware RAID card. If you're looking at say more than 4-5 drives worth of storage capacity, you need 3-4 more HDDs in a duplication setup vs RAID and that can easily outprice an 8-port RAID card. Not to mention the difficulties housing and powering >10 drives.

Oh, just thought of one: RAID handled your back-up in real time. With the external drive approach the data between scheduled back-ups is at risk.

Just to beat the horse one more time though, RAID isn't backup nor is it a substitute for backup, RAID is redundancy. I think what most of us here use RAID for is twofold, one it's handy to have everything on one volume, but moreso it's to save us from having to re-rip out music and movies if a drive fails. It's not really to "back up" the data since that's already done by us having the original discs.

[tchapin]

Do both. The additional cost of an external USB HD is relatively minor.

Depending on your computer's case, you may or may not want to get a RAID cage to help make it easier to install / replace the drives, and to save space inside, if you have a bunch of 5.25 bays open.

[GKusnick]

An alternative to an external drive is a regular internal SATA drive in a hot-swap tray (assuming you have front panel space for it). Such trays are fairly inexpensive and the total cost might be less than that of an external drive.

Either way, I'd suggest you get two backup drives instead of just one, and alternate them on a daily or weekly basis. The one that's not in use at any given time should ideally be kept offsite, or if that's not practical, in a sturdy fireproof box in your basement so it has some hope of surviving a disaster.

[davephan]

It sounds like you are trying to develop a disaster recovery plan, which is a very good idea. You should have a recovery plan before you need it, so you can avoid a time consuming and painful manual system rebuild with a lengthy SageTV outage. As mentioned by others, RAID will help keep your system running if a drive fails. You can confidently run a system with out RAID if you have a recovery plain in place.

What you really need is disk imaging software, if you are not using Windows Home Server for your operating system. I haven't found disk imaging software that can recover the WHS operating system. You can easily and quickly recover your SageTV computer back to a point in time before the failure occurred. The disk imaging can really reduce downtime and improve the WAF.

I use Ghost now. I might switch to Acronis in the future.


Dave

[Greg]

Thanks to all of you for the discussion. So much to learn.........but I'm enjoying it. If there's one thing I learned it's RAID is not a back-up!

It does sound like RAID is a good idea, as well as a back-up drive.

Any reason NOT to use RAID?

Originally Posted by davephan

What you really need is disk imaging software, if you are not using Windows Home Server for your operating system. I haven't found disk imaging software that can recover the WHS operating system. You can easily and quickly recover your SageTV computer back to a point in time before the failure occurred. The disk imaging can really reduce downtime and improve the WAF.

I use Ghost now. I might switch to Acronis in the future.

Dave

Thanks for the tip. How often do you take an image? Is the process pretty simple?


I'll be asking more computer build/component selection questions as I take this path.

Thanks for the help.

[stanger89]

Originally Posted by Greg

Thanks to all of you for the discussion. So much to learn.........but I'm enjoying it. If there's one thing I learned it's RAID is not a back-up!

It does sound like RAID is a good idea, as well as a back-up drive.

Any reason NOT to use RAID?

What are you actually trying to do? Is it a full system disaster recovery plan, or is it just trying to have some level of "insurance" against re-ripping digital (but recoverable) media?

[davephan]

Originally Posted by Greg

Thanks to all of you for the discussion. So much to learn.........but I'm enjoying it. If there's one thing I learned it's RAID is not a back-up!

It does sound like RAID is a good idea, as well as a back-up drive.

Any reason NOT to use RAID?



Thanks for the tip. How often do you take an image? Is the process pretty simple?


I'll be asking more computer build/component selection questions as I take this path.

Thanks for the help.

I would like to take an image at least once a month or more frequently. However, I sometimes go for several months. It is possible to setup imaging to do incremental images much more frequently with a scheduled job. If you were imaging very frequently, then you could recover to a few minutes in the past. Generally though you want to recover to some time in the past to ensure that your system was still working properly and the problem was not in the developing stages.

Recently, I had a problem with my HDHomerun unit. I could not get the HDHomerun unit to work properly after deleting the tuners, re-doing the drivers, etc. The HDHomerun was working fine with the free Hauppauge software, but the channels were not detected by SageTV. I had an image that was about a month old when everything was working perfectly. I recovered back to that C drive image. Now everything works perfectly again, and has remained that way for a couple days (and counting) after the recovery.

From time to time, I do have trouble with the SageTV server, but for the most part it is reliable. I don't know what caused the problem, but whatever caused the problem doesn't matter because the problem is fixed, and appears to be staying fixed. It would have been a pain without the image and more SageTV downtime.

RAID is a good idea. I don't know for sure if it works with WHS. The downside to RAID is you can't recover back in time when the system was working perfectly. Do do that you still need images. However, it does protect you if a hard drive fails. But images also protect you from a hard drive failure too, except for the changes since the last image would be lost.

I did have to re-do many configuration changes for my new HD-200 after I recovered with the image. I then took another image after a couple days of solid operation to capture the changes since the last image a month ago.

Dave

[jsonnabend]

Originally Posted by GKusnick

I still wouldn't consider that a backup since it doesn't save you from needing a real backup solution. What it does save you is the need to restore from backup in the common case of single drive failure. RAID's data redundancy is how it achieves its function of minimizing downtime, not a true backup function.

Well, that's a bit of semantics. It is definitely NOT a true backup, but when it functions properly, it protects against data loss for a single drive failure.

Originally Posted by Taddeusz

Not true. If you have a lot of static files, as is the case of many of our SageTV setups, bit rot can creep into the parity data of a RAID 5 array or into the drives of a RAID 1 array. Your RAID rebuild will fail in these cases.

Or, depending on the size of your array, a second drive failure during rebuild will take down the whole array. After a single drive failure the likelihood of a second drive failure goes up quite a bit.

Again, RAID is not a backup.

Oh please. Yes, if all hell breaks loose, then RAID is for sh!t, both as backup and to minimize downtime. Same could be said for your precious backups, though, too.

I had a drive fail in my raid 5 box. I brought down the server, popped in new drive. Fired up the server and had the array rebuilt in a couple of hours. A failure of a single drive didn't result in data loss, and that's what most people use backups for.

- Jeff

[sic0048]

Honest question here - what about some sort of corruption event. Say a power flicker that somehow causes a bad write on the systems disk. If you are mirroring the system drive, would the corruption occur on both disks?

I've always wondered about that. I would consider mirroring my WHS systems disk (if that is even possible using a hardware RAID card), but wondered if it would really do any good (other than protecting against a simple 1 disk failure).

[stanger89]

Originally Posted by jsonnabend

Well, that's a bit of semantics. It is definitely NOT a true backup, but when it functions properly, it protects against data loss for a single drive failure.

It's semantics but semantics are important. Most people probably understand the concept that they should "back up" their important data, whether they actually do it or not. They understand the idea that you need a separate copy of the data in case any number of bad things happen to the PC (virus, hardware failures, software failure, user failure, environmental failure, etc).

Calling RAID backup can easily give people the idea that it's a valid solution for protecting data against all of the above like a real backup solution would, when in fact it doesn't, it only protects against one of those, and then only a specific case of it (disk failure).

A backup strategy is essential if you have important, unrecoverable data. What it really comes down to IMO, is RAID is not a necessary part of a backup strategy, and even if you include it, you need other measures as well.

Thus RAID cannot be considered a backup solution, unlike duplication to disconnected drives, tape backup, burning DVDs, etc.

[tchapin]

Originally Posted by davephan

I use Ghost now. I might switch to Acronis in the future.


Dave

Hi Dave,

I'm interested not necessarily in imaging a drive, but cloning it. If my system drive fails, I'd like to be able to take the spare drive that I've been cloning the system drive on to and replacing the dead drive in my system with it.

Does Ghost do that?

Acronis doesn't. It takes images, but they are in a proprietary Acronis format. I use their True Image product at work. To restore, you need to boot into a special mode and restore the image onto the new drive. So, effectively, you'd need three drives 1) the original 2) the one holding the image and 3) the replacement.

I'd like to be able to combine 2 and 3.

Thanks.

Todd

[gplasky]

Any imaging product makes a single image file for you to resotre to another drive. I'm not aware of any imaging product that makes a second bootable drive for you. What you want to do is mirror your OS. Either by RAID hardware controller or thru a software RAID mirror. Or if you do RAID you want the drive to be a hot spare.

Gerry

[stanger89]

Originally Posted by sic0048

Honest question here - what about some sort of corruption event. Say a power flicker that somehow causes a bad write on the systems disk. If you are mirroring the system drive, would the corruption occur on both disks?

Yes exactly. If the power flickers and corrupts the drive, it will (probably) corrupt both, if you delete a file, gone on both, if a virus infects you, it infects both, etc, etc. A big reason why we say RAID isn't backup.

[Peter_h]

I can attest to this. I had this exact scenario happen with my WHS recently.

RAID 1 on the OS, a nice power flicker, data corruption and an OS that wouldn't boot past the windows is loading screen.