How many spares do you use in your RAID set?

W

war9200

Limp Gawd
Joined
Oct 9, 2010
Messages
273
Hi,

For a RAID-X where X is any RAID, 1,3,5,5,10,50,60 etc, (not judging RAID type here), how many spares do you use?

I currently use 1 in a 15TB RAID-6 (+1) = 16 drives total.
Does it make sense to use 2 if using ~19 drives in a RAID-6?
 
Do you think 2 drives will die in the same day?

I generally use 1 spare. And sometimes share that between more than 1 array.
 
drescherjm said:
Do you think 2 drives will die in the same day?

I generally use 1 spare. And sometimes share that between more than 1 array.
Click to expand...

If RAID6 is used, then you're good.

At my job, we keep two-three hot spares ready at all times for the servers, and our SAN.

For non-mission critical stuff, hot spares are icing on the cake, but for anything that needs to be protected, they are a must.
 
0 hot spares, don't want to waste a port on a spare, but I always keep a cold spare.
 
At my company we generally recommend 1 Hot Spare for every 30 drives regardless of RAID type. The systems we sell are pretty high end, however and we have 7 x 24 x 365 warranty coverage on all disks with a 4 hour response time and extensive pre-failure analysis. In my house I don't use spares, since I can shut off the system in the event of a failure and replace the drive at my leisure. The question is, how long can YOU afford to wait for a replacement? If you don't have a great support group swapping drives, and the data is critical, then I would say 2 at a minimum for 1-30 drives and then 1 per 15 after that, depending on how big the overall system is. That way, you can have one or two drive die in a couple of sets and you are still ok.
 
For my Raid 6 array, I intend to have 1 hot spare.

I rarely turn my desktop off, and if a drive goes bad, I'd like for the Raid card to take care of the problem immediatly. When I come back to my computer the next day, it'd be nice to see that even though a drive failed during the night, the array was already well on the way to rebuilding parity on the hot spare.
 
At work I have nagios monitoring all machines, arrays and the SMART for every disk in all of the servers. With the smart data I can usually predict that a drive is going bad and swap it out before it gets kicked out of the array. I have > 50 bad sectors and also > 0 Offline Uncorrectable or > 0 Pending sectors all being CRITICAL alerts in nagios.

http://www.nagios.org/

I do not immediately swap out drives that have > 50 bad sectors. I watch them daily on my nagois monitor and if the numbers increase daily or weekly I take action. Currently out of my 15 arrays there are 3 drives that have > 50 bad sectors but none of them are growing.
 
> Currently out of my 15 arrays there are 3 drives that have > 50 bad sectors but none of them are growing.

Do you run regular scrubbings?
What happens if you have a disk failure/rebuild one may fail?
 
Not regular scrubbings but I plan to enable this. When I have done scrubbing I have not seen any problems. Although I was confusing I am talking about the SMART value of "Reallocated Sectors". Which means these are the sectors that the drive was able to read and it successfully reallocated the weak sector to a new sector from the spare pool.
 
What happens if you have a disk failure/rebuild one may fail?
Click to expand...

I have never experienced a disk failure during a rebuild in 15 years of running raid 5 and 6. Although I have had at 20+ disks go bad in arrays. I catch most of them now before they die.
 
You must log in or register to reply here.
Back
Top