For you IT guys: I have 2 raid-1 sets w/ an adaptec controller in my wife's office server. Been there since I built the computer in 2004, it runs 24/5, sometimes 24/7. I am thinking of proactively replacing the HDs one at a time and let it rebuild, then the other. If I wait until one breaks, it will be expensive in terms of down time. I plan on replacing the system disks (74GB Raptors) and data set (160GB WD) with 500GB WD Blacks. Is this asking for trouble? Also when changing out the first one, will the difference in size cause any problems?
Mike

If this is your only kind of back up you also need to think about off site storage of some kind.
What type of down time and loss of service will a fire bring? Water damage from a fire in a accompanying office or floor above you?
For the RAID I would be careful and do as much research as possible into the storage devices you have. I have heard many horror stories about doing what you are planning or something similar and losing a significant amount of time rebuilding.

Regardless, have a plan in place just in case a worst case scenario happens.

I would back it up to tape or some other media and leave it alone. FYI, if you are trying to gain disk space you will not with this method. The RAID will stay the same size as it was created.

I'd do a complete backup before you start pulling hard drives, moving things around, etc. Also, verify you can get data off that backup.

I second all of SuperDave's concerns. RAID-1 doesn't protect you from user error, disaster or theft.

What down time would you have if one hard drive breaks? 0. So I'm not sure why you would do this.

If you're looking to add space then add another controller and RAID set or use one of those cheap home network NAS solutions that support RAID.

I think what the OP is getting at is he's interested in replacing the hard drives because they are 5 years old. There's no reason why his method shouldn't work. That said, I think it would be money well spent to pick up an external USB drive of sufficient capacity to do a full backup before fiddling with the server. That will get you a usable backup and you can then put that backup offsite if the mood strikes you. Murphy tends to show up at the most inopportune times so a backup would be my first order of business.

Also, for what it's worth, I've got some 9 year old quantum fireball 20gig hard drives that have been running 24/7 in a test server since 2000. I've also seen brand new hard drives fail within minutes of being installed.

Why would you have any expensive downtime if one breaks? Wouldn't you just run on the single drive until you replace the other drive?

I do agree with having a backup made, even with new drives.

I'm using Carbonite for offsite backup. A friend had his office computer go down and darn if the Carbonite didn't come through. So I'm satisfied that it will work. Familyguy is right on, as to my intention. The girls in the office never look at the server and wouldn't know if one of the drives failed. Actually I don't know what would happen; after the boot when its running on one drive would Windows pop up a message? At any rate I was wondering if anyone did a proactive replacement or just waited until one died. Sounds like the consensus is to let them run. I'm not trying to increase the size of the installed raid setup; but the 500MB Black is the smallest black. I'll be happy to replace the Raptors; they're the noisiest drives I've ever used.

With older gear, it seems the best policy is don't mess with it.

When does stuff fail? One of the most frequent failure scenarios is when it's powered up.

So a drive can run for years and you lose power, and it doesn't spin back up.

We just went through this a year ago when a major customer lost power to their downtown data center due to a water main break and the water coming into the power switch panel through the conduit that was feeding one of the power sources.

So all the automated power switching, batteries, etc were no good because the gear flooded.

Anyway, we had 3 or 4 days of 24 hours on site getting equipment up and running again as drives failed to spin up after being powered off.

Systems that ran since 1997 were losing drives and power supplies.

So stuff will run, until something happens.

I see the same sort of thing when people want to upgrade older gear. It's nerve racking to power it off because one doesn't know if it will power back on, or we create more problems, etc.

We do sanity power cycles and reboots on major service actions just to make sure the system will reboot cleanly before we make any changes.

So I'd add to the plan to add a USB or some sort of drive before you crack the case, because things may not appear.

Also, I really doubt you can remove one drive and mirror that drive to a new drive. I think you would have to add the two new drives, mirror them and then copy the data to the new drives.

But check with the RAID controller docs to see what can be done.