Ben Russo said: > Really? I haven't had the fun of playing with any IDE HW Raid devices.. > But do you have LOTS of experiences with this? Or is it just a few cases? > I am curious because I am considering purchasing some Promise HW IDE RAID > arrays that have SCSI Host Interfaces? They just seem so damn cheap.
well I can share some of my experiences. I'm sure they are somewhat biased but I've done a lot of research on it... About a year and a half ago I deployed 3 hardware raid 5 systems, each had 4 75GB IBM 75GXP disks in hardware raid 5 on 3ware 6800 8-port IDE controllers, and 1 30GB IBM disk connected to an intel ide controller (MB was intel L440GX+). Systems were single proc P3-800 with 256MB ECC ram, 300 watt power supplies, 4U extended ATX chassis with 3 large fans, not sure exact size but I would guess 5"x5". anyways, over the span of 6 months I would estimate we had about 10 disk failures on these systems alone. on one unlucky weekend we had 4 disks fail accross 2 systems in 3 days, wiping out the entire raid array in the process on one of the systems(the other system, the 2nd disk that had failed was a brand new one that was fedex'd overnight for saturday delivery, packed in the standard 2-3" of foam, it was DOA(bad sectors on arrival, the controller refused to use it). We replaced the disk coolers, power supplies, the disks themselves, finally settled on 6 x 80GB maxtor drives in raid 10, with a 30GB maxtor drive as the boot drive for each system, 450watt power supplies, failure rates went way down, but we still had I think 3 disk failures in the past year(far better then what IBM gave us). A 4th disk is on the verge of failure. For the longest time I placed blame on these systems, I thought perhaps it was the power supply, it was the raid card .. but as time went on, I had disks(these were all IBMs, at the time I had everything IBM, since I had such good experience in the past with their 10-36GB drives) fail in sun ultra 10s, P3s, P4s, other dual proc p3s, home systems, work systems .. co worker's home systems.. It was a distinct pattern. and of course theres a lawsuit against IBM for the drives(which I'm one of the main people involved). Maxtor has been much better, though as above they are not perfect. I bought a 100GB maxtor drive around aguust 2001, the drive sat in a very well run setup, the drive was cool or cold to the touch at all times, connected to a good battery backup, on a 300 to 450 watt power supply(I upgraded halfway through). In april of this year the drive hiccuped and I lost some data(pissed me off the most when all my MRTG logs were gone). It continued working, I migrated again this august to my new western digital 8MB cache raid1 setup, and retired the maxtor to inactive duty, shortly after it started spitting out errors(kernel would say device is not ready even though it wasn't even mounted). So I convinced maxtor to RMA it, reciveed the new drive, haven't had a chance to try it yet. But I did have at least 3 80GB maxtor drives fail in my 3ware systems, and 1 more on the fritz, it acted like it failed 2 or 3 times but has recovered every time sofar and the last remaining IT guy at the company is too busy to worry about it, the system it's on is a backup server, so if it were to lose all it's data it wouldn't be a big loss, just a pain to rebuild everything. I've read also that many IDE manufacturers have reduced their warranties to 1 year, with the exception of a few models like my WD special editions which I hear have a better warranty.. But to be honest I have lost more drives(most of which are IDE) in the past 2 years then I have since I started using computers back in around '91. I can count the # of failed drives before that probably on one hand. It's quite scary to me. I mean I shouldn't have to resort to raid to maintain data integrity. I mean, a drive life of 2 years or 3 years would be fine, but having drives die before even 1 year is up is just plain horrible. And having replacement drives die within months of replacing them is even worse. much of my failures occured in a climate controlled enviornment, with good quality 1.4KVA battery backup systems ..so it's not like I treated them badly, they had the best care of all of the systems on the network and they still had massive failures. I literally spent several days worth of time replacing disks, rebuilding systems and reinstalling them. our backup servers for example, it could take up to 2 hours to replace a disk, they were real hard to work with, having so many ide cables and power cables in such a small spot, and me having big hands, doesn't make for easy drive replacement :( And I have heard stories from others as well, so I know it's not me, I remember last year on the phone with 3ware, they specifically told me they had customers returning literally thousands of IBM 75GXP drives due to failures. I am happy sofar with my WD special edition drives, but I only got them a few months ago, who knows what they will be like in a year. meanwhile I have several SCSI drives that have been humming for years and years without the slightest problem. A couple of my SCSI controllers are more then 6 years old :) still use them everyday .. my experiences are probably biased but I've had some real bad luck with IDE drives. IDE is still good for big storage at low cost.. but I would reccomend SCSI over IDE for reliability's sake. I haven't used many HP disks myself(aside from ones in my HPUX machines which are probably HP). Most of the SCSI disks I deployed were IBM as well, with a few being quantum(now maxtor I think), and a few seagates. But the vast majority of the SCSI disks were IBM. Much of my home equipment is IBM disks too, since I haven't bought a new SCSI disk in ages, most of what I bought was used..but still no troubles. nate -- redhat-list mailing list unsubscribe mailto:[EMAIL PROTECTED]?subject=unsubscribe https://listman.redhat.com/mailman/listinfo/redhat-list
