RAID Risks

[Sovereign]

I know that "You should never store any data you really care about on a RAID0 array without regular backups" and "For every drive you add to a RAID0 array, you are increasing the risk of array failure."

Given that I do not require a re-hashing of the traditional risks of RAID0, is the inherent risk really that much higher if I added another 250GB WD2500KS to my existing setup? The 500GB array is getting a little cramped right about now and I was considering adding more space this summer. I use the onboard nForce controller (which I'm sure someone will take their time to bash/diss as being a piece of crap) which has never caused me any problems at all.

My drives are not exposed to excessive heat or overclocking (the 2.6GHz OC is incorrect, I will have to change that, because the Opteron 170 is currently running at 2.0GHz).

 

[drizzt81]

Given that I do not require a re-hashing of the traditional risks of RAID0, is the inherent risk really that much higher if I added another 250GB WD2500KS to my existing setup?

That much higher than... ? What is the alternate option?

Looking at your sig, I am guessing that you'd like to move from a 2-disk RAID-0 array to a 3-disk one. Why not get a third drive and mount it to a folder? Seeing that 320GB or even 500GB drives are a better bang/ buck than 350GB ones, it's what I would do.

Data loss risk:
If you run a 3-disk raid-0 array, the risk of ANY data loss due to drive failure is equal to the risk of ALL data loss due to drive failure. If you add a third drive that is not in the array, the likelihood of any data loss is equal to the previous probability, but the chance to lose all data is less.

 

[Xeth]

You'll get a ~50% increase in risk. According to the Google study the average hard drive failure rate is around 5% so you're going from ~10% to ~15% chance...

Of course that's only if you run the drives 24/7...

 

[Sovereign]

They are run 24/7 but they are not being accessed 24/7. However, if the "failure" you're speaking of refers to physical failures such as drive motors (rather than logical failures, i.e. controller errors), then there is no difference...

I didn't know I could get another and mount it as a folder. Does this mean that the drive acts like a "folder" on the RAID array (that is, I only see two drives) but it acts like the RAID array has been expanded? Or do I see three but just have a convenient way to drop things into the third one?

 

[GN85]

You'll get a ~50% increase in risk. According to the Google study the average hard drive failure rate is around 5% so you're going from ~10% to ~15% chance...

Of course that's only if you run the drives 24/7...

Also depends on how old his 2 hard drives are.

If you were to install 3 new hard drives at the same time, there would be a bigger chance of one failing than if you add a 3rd drive to 2 that have been running for a year and have proven to be free of any manufacturing defect.

Hard drives typically fail within a short time if there is a defect present from manufacturing. If a hard drive proves reliable for a period of a few months after installation, the hard drive has a significantly greater chance of remaining reliable.

 

[unhappy_mage]

I didn't know I could get another and mount it as a folder. Does this mean that the drive acts like a "folder" on the RAID array (that is, I only see two drives) but it acts like the RAID array has been expanded? Or do I see three but just have a convenient way to drop things into the third one?

I'm not sure what you mean by either option. Here's what'd happen, though. If your current array shows up as the J: drive, you could tell Windows to mount the new drive on J:\newstuff. Then anything you put in that folder will be stored on the new drive instead of the array.

 

[Sovereign]

The two drives I have were run 24/7 starting September last year. I purchased them in May of last year. No defects or errors have shown up since then.

 

[drizzt81]

The two drives I have were run 24/7 starting September last year. I purchased them in May of last year. No defects or errors have shown up since then.

That is nice. The google study claims that SMART is pretty much worthless at predicting drive failures. I don't recall whether there were any statistics about when drives failed in that study.

However, if the "failure" you're speaking of refers to physical failures such as drive motors (rather than logical failures, i.e. controller errors), then there is no difference...

IIRC the study meant physical failures. You can read it here:
http://216.239.37.132/papers/disk_failures.pdf

I didn't know I could get another and mount it as a folder. Does this mean that the drive acts like a "folder" on the RAID array (that is, I only see two drives) but it acts like the RAID array has been expanded? Or do I see three but just have a convenient way to drop things into the third one?

It depends how you'd like it to act. In WinXP Pro you could have it mounted as a folder and have it assigned a drive letter. Either way it separates the file systems between the new drive and the two old ones, so that a HDD failure in your RAID array would not destroy the data on the separate drive and vice versa.
Using folder paths (I think that is the term for it) does reduce your flexibility, since your space is partitioned and not a single, large area.

 

[Sovereign]

I am thinking about adding a large (400GB+) drive to my collection, or just getting an external (so my movies and music become portable)...