Hot Spare or Not Hot Spare

[scott_417]

Assume I built an array with 4 disks in it and RAID 5 implemented.

1) Use One Disk as Hot Spare

Advantage: Obvious
Disadvnatage: Lost one disk worth of storage capacity

2) Use NO Hot Spare

Disadvantage: Obvious
Advantage: Gain one disk worth of storage capacity

Now, If I do not use the hot spare, instead put all 4 disks under the RAID 5, and if one disk went bad:

3) What is the easiest way to detect it on a day to day operating basis?

4) If I can easily detect the bad disk, can I just PHYSICALLY swap it with a new disk of the same capacity?

5) If YES to 4), it seems to me use NO hot spare could be more efficiently take good use of the disk capacity. But please do share you experience, knowledge and comments.


Thanks.

[Miika T · Accepted solution]

3) Depending on the amount of servers you have, HP Systems insight manager might do the trick. It would send you and alert, if a fan, disk, etc. dies.

4) Yes, smart raid will start rebuilding the array immidiately.

5) Various factors should be considered here. If you actually monitor the servers and have sim to send you alerts, maybe you won't need a spare, you should always get a new drive the next business day atleast if the server is in warranty. If you do all admin stuff yourself, maybe more than one drive will fail during your vacation....

-Miika

[David Claypool]

3) What is the easiest way to detect it on a day to day operating basis?

**The ProLiant Support Pack and HP Systems Insigth Manager

4) If I can easily detect the bad disk, can I just PHYSICALLY swap it with a new disk of the same capacity?

**Yes (as long as we're talking about a ProLiant 300/500/700 or BL series and a SmartArray controller)

5) If YES to 4), it seems to me use NO hot spare could be more efficiently take good use of the disk capacity. But please do share you experience, knowledge and comments.

**During the time that you are operating after the failure and before the disk is replaced, you will critically fail if another drive fails. Usage of a hot spare is usually decided on the criticality of the system in question and the availability of personnel and parts to make the replacement.

Example: Major chain retail store using server for store operations. Spare parts are not stocked at the location and personnel are not educated on the systems. Next business day service contract.

Clearly a case for hot spare to eliminate the danger of a failure in the interim.

[Shameer.V.A]

Hi Scott,

3) What is the easiest way to detect it on a day to day operating basis?

If Monitoring softwares installed, it will tell you. The fail LED on the purticular disk will be active. RAID controller will give you alarm (depends on model)

4) If I can easily detect the bad disk, can I just PHYSICALLY swap it with a new disk of the same capacity?

Yes , You can just insert the new one ( Hot swap HDD) same or higher capacity can be used ( RPM is not must to be equivalent, though it will affect the performance slightly)

5) If YES to 4), it seems to me use NO hot spare could be more efficiently take good use of the disk capacity. But please do share you experience, knowledge and comments.

If daily monitoring is there, then it is not must to use hot spare. But now disks are so cheap, worth to take precautions.

"PLAN for the WORST, & HOPE For the BEST"

Regards,

Shameer

[scott_417]

Thank you all for the responses.


1)When online spare is employed, in the case that one disk in an array goes bad, the hot disk will be built into the array. But, after the hot spare is put into operation, there will be no hot spare any more. So I have to MANUALLY replace the bad disk with a good one, in order to allow the system to return the Hot Disk to its orginal online spare mode and ready for next crash.

2)Now, as compared to the above case, I do not use the hot spare, instead, put all available disks into operation. In the case that one disk goes bad, I'll MANUALLY swap the bad disk with the good one.

As summary, in both Case 1 (hot spare) and Case 2 (no hot spare), I'll have to MANUALLY do a disk swap (once) anyway. I just wonder why should exercize the hot spare strategy?

After the hot spare, if I do not repalce the bad disk with a good one in time, during this period, if another disk goes bad in the array (those bad days and bad time), the system will be left running without any spare disk. In another word: even if the hot spare strategy is employed, I have to monitor the system in a daily basis, and be ready to do the MANUAL swap. This is equivalent to the one without using the hot spare strategy, just lost one-disk-worth capacity.

Please share your experience and comments.


Thanks

[Shameer.V.A]

Hi Scott,
Hot spare will auto include the failed disk position in the array. & When you replace the failed disk, you can either continue with the same setup & make replaced disk as new hot spare or you can go to the old setup ( by rebuilding the array again to the replaced disk & use old hot spare disk as hot spare again).

When you are having more than one array, in that case you can have a "global hotspare", which will use a single hotspare disk for more than one array . Only thing you should use the higher capacity disk in the array as hot spare ( eg. If you have an array of 146GB & array of 300 Gb, you should use 1 300Gb as a hotspare).

As per your summery..
Hot spare is meant for additional data protection, by rebuilding the array as soon as one disk in the array is failed. So that if one more disk failure after the rebuild operation also can be taulerated.

In a normal RAID 5, if one disk is failed, the performance also will be very worse. &one more disk failure can lead you to total data loss.

Hope this will help you to conclude..

Regards,

Shameer

[Shameer.V.A]

Hi scott,

Please close the thread, if your query is completed..

shameer