A few comments since this is something I do.
We have 11 Dell servers and have been very happy with them. We've had no trouble with them at all. Also, if you get them on the phone you can negotiate the price down. I've had them match memory prices with the best I could find on the web. Recently we found a local vendor for servers that can get our Sun and Dell gear for a better price than I've ever been able to get.
Also, one thing to keep in mind with regards to the extra expense with Dell or Compaq (or Sun, IBM, etc. ) is the service contract. We get 4-hour service contracts for our datacenter-based servers (they usually include next business day for no cost). This is very valuable--I've not had to use it except once (just recently actually) on a Dell laptop and it worked as advertised. They'll send someone out to replace hardware on 4-hour notice any time of the day--and it seems servers always konk out at 3 in the morning.
Now, for a backup machine I do have a Tyan-based Athlon 1600+ MP system with a 3Ware IDE RAID controller and 6 Maxtor 100-GB drives in 3Ware hot-swap drive trays. Most excellent system. I have had one drive fail--I'm running it in RAID 5 with one hot spare--but it was replaced under Maxtor's warranty. For this machine RAID 5 is fine but for a production machine by all means use RAID 10. Use good cooling fans on the CPU though and make sure you have enough power. During testing I started the machine up, put Seti@home on it and watched the load of the machine rise until - poof - the circuit breaker on my 650 VA UPS blew. Man, those things do draw the power. I'm using a Lian-Li server case and won't use anything else from now on for machines I build myself.
BTW, most of our machine run RedHat 7. 2--some of the older ones run are on 6. 2 and a few on FreeBSD. I had the dual Athlon on Suse for awhile but I couldn't get used to it and put Red Hat on it (it has a SCSI boot disk so I can do this without disturbing the data on the RAID array).
BTW, with regards to SCSI drives: for years I've not had a disk fail until the past few months. I upgraded one of our older Dell servers with 4 IBM 15k RPM 160 drives and had two of them fail within 2 weeks. The hard drive on the above Dell laptop (also fairly new) was IBM (and replaced by Dell with a Hitachi) and I've had an IBM drive in a Sun die, too. Bad luck.
Since I had replaced the drives in the Dell server myself the support contract didn't apply--I didn't have two spares of this size disk on hand and my customer on that machine wasn't willing to wait 14 days for IBM to replace it under warranty--so I ordered new Seagate 15k RPM SCSI 320 drives. Perfect. And much better performance than I was getting with the IBMs, too.
Also on the Dell servers--it's true you can't just yank disks out of a running machine. But you can use the afacli utility to remove a disk from the RAID container then remove the drive, replace it, and rebuild the array--all while the machine is running--at least on Linux and FreeBSD. I've tested this--it works. You might have to open the case, though--our 2U servers have a "pin" though the drive cages that have to be removed before you can release the drives--I normally leave the "pins" out so I won't have to mess with them if a drive fails.
Attention: TDR Forum Junkies 
