Raid Question

nowhereman · 08-18-2011, 02:04 PM

One of my external drives has begun to delete images randomly - there one day, gone the next time I need them (of course). I'm looking into a Raid setup, (multiple dedundant backups) and wonder if anyone has experience with them or could make reccommendations as to a decent setup. Thanks!

Undertoad · 08-19-2011, 11:26 AM

My experience with them is "set it and forget it" but there are three pieces of advice:

1) RAID 1, not RAID 5. Apparently there are issues with modern hard drives where the drives take too long to recover in a RAID 5 situation. I don't know the details of this problem but my approach is allways KISS, and RAID 1, where two drives just become identical copies, seems like the simplest approach. This way if one drive fails, the other takes over; and, the best part, one drive can be taken out and run outside a RAID if need be. If you take out one RAID 5 disk it is useless.

2) Hardware RAID, never software. A motherboard that can do hardware RAID is only like $10 more.

3) Never RAID 0. The zero in 0 is the number of files you will recover in case of failure.

Flint · 08-19-2011, 04:05 PM

Raid 0 is a meme that says "yo dawg I heard u don't like redundancy so I didn't put any redundancy in ur redundant array"

Nowhereman, Jeremy, Hillary, PHd, have you considered an online backup service, or are you just more of a do it yourselfer?

mbpark · 08-20-2011, 09:19 PM

Ut,

We run RAID 5 on hundreds of servers and an EMC Clariion SAN (think about $2M) with over 20 VSphere boxes and 500TB of storage. We run it because it allows us to keep drive arrays running while replacing bad drives. Your statement there is not true for servers and other hardware. In many cases, we run RAID 6 so we can have two drives fail and still keep the array (my smtp bridgeheads run this).

The only time in the past five years I have had an issue is because the quality of the drive array controllers in HP servers has gone to the toilet in a few cases. We had bad array controller corrupt NTFS during a rebuild on a Proliant DL 380 G5. We had to restore from backup and regenerate the rest of the data. The other 99.5 percent of the time we just replace the drive and there is no issue. This includes servers from Dell, IBM, HP, and even some whiteboxes.

And yes, I would recommend RAID 5 and a spare drive in a heartbeat if you have a decent controller. RAID 1 is more cost effective however.

mbpark · 08-20-2011, 09:21 PM

Btw i have rebuilt raid 5 and 6 arrays with modern SAS drives in under 2 hours. It's way quicker now with modern controllers. Just did 2 a couple months back, one on a sql server, and the customer barely noticed.

08-18-2011, 02:04 PM	#1
nowhereman sliding down the razor blade of life Join Date: Jan 2007 Location: just over the edge Posts: 228	Raid Question One of my external drives has begun to delete images randomly - there one day, gone the next time I need them (of course). I'm looking into a Raid setup, (multiple dedundant backups) and wonder if anyone has experience with them or could make reccommendations as to a decent setup. Thanks!

08-19-2011, 04:05 PM	#3
Flint Snowflake Join Date: Mar 2006 Location: Dystopia Posts: 13,136	Raid 0 is a meme that says "yo dawg I heard u don't like redundancy so I didn't put any redundancy in ur redundant array" Nowhereman, Jeremy, Hillary, PHd, have you considered an online backup service, or are you just more of a do it yourselfer? __________________ **************** There's a level of facility that everyone needs to accomplish, and from there it's a matter of deciding for yourself how important ultra-facility is to your expression. ... I found, like Joseph Campbell said, if you just follow whatever gives you a little joy or excitement or awe, then you're on the right track. . . . . . . . . . . . . . . . . . . . . . . . . . . Terry Bozzio**

08-20-2011, 09:19 PM	#4
mbpark Lecturer Join Date: Jan 2001 Location: Carmel, Indiana Posts: 761	Raid 5 Ut, We run RAID 5 on hundreds of servers and an EMC Clariion SAN (think about $2M) with over 20 VSphere boxes and 500TB of storage. We run it because it allows us to keep drive arrays running while replacing bad drives. Your statement there is not true for servers and other hardware. In many cases, we run RAID 6 so we can have two drives fail and still keep the array (my smtp bridgeheads run this). The only time in the past five years I have had an issue is because the quality of the drive array controllers in HP servers has gone to the toilet in a few cases. We had bad array controller corrupt NTFS during a rebuild on a Proliant DL 380 G5. We had to restore from backup and regenerate the rest of the data. The other 99.5 percent of the time we just replace the drive and there is no issue. This includes servers from Dell, IBM, HP, and even some whiteboxes. And yes, I would recommend RAID 5 and a spare drive in a heartbeat if you have a decent controller. RAID 1 is more cost effective however.

08-20-2011, 09:21 PM	#5
mbpark Lecturer Join Date: Jan 2001 Location: Carmel, Indiana Posts: 761	Rebuild time Btw i have rebuilt raid 5 and 6 arrays with modern SAS drives in under 2 hours. It's way quicker now with modern controllers. Just did 2 a couple months back, one on a sql server, and the customer barely noticed.

08-19-2011, 11:26 AM	#2
Undertoad Radical Centrist Join Date: Jan 2001 Location: Cottage of Prussia Posts: 31,423	My experience with them is "set it and forget it" but there are three pieces of advice: 1) RAID 1, not RAID 5. Apparently there are issues with modern hard drives where the drives take too long to recover in a RAID 5 situation. I don't know the details of this problem but my approach is allways KISS, and RAID 1, where two drives just become identical copies, seems like the simplest approach. This way if one drive fails, the other takes over; and, the best part, one drive can be taken out and run outside a RAID if need be. If you take out one RAID 5 disk it is useless. 2) Hardware RAID, never software. A motherboard that can do hardware RAID is only like $10 more. 3) Never RAID 0. The zero in 0 is the number of files you will recover in case of failure.

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)