Questions on MSA1000 drive array rebuilds/transitions

[Cameron Todd]

I have a 6 spindle RAID5 array defined, with several logical volumes that are in use.

Is it sensible, if indeed possible, to be able to convert this array from RAID5 to RAID1+0 while on-line and servicing connected hosts? (The option to select RAID type seems to be available when viewing existing arrays in the Array Config Utility but I wouldn't dare try it untested)

Also, correct me if I am wrong, but I assume that, despite having ample free space, it is not possible to reduce a 6 disk array to 5 disks on the fly.

One last query. Is there a utility available that can test the data throughput performance of the MSA system, as I have been trying to measure different disk configurations (varying stripe sizes, spindle numbers, array locations) using a defined amount of file data (a few CD's worth in fact) and a stopwatch, and the results are a bit vague given that there are several different components in the chain that could be acting as a bottleneck for each test.

[CA1309061]

Hi Cameron!

You can build RAID 1+0 Array only on even number of physical disk. You may online convert/migrate your RAID (defined Rebuild priority on MSA controller settings) if your have enough free space beyond of current Array.

You may try IO meter utility for testing purpose http://www.iometer.org/

[Cameron Todd]

Yes, I discovered that did need an even number of spindles. I also discovered via this A-B testing that RAID 1+0 wasn't significantly faster than RAID5 (reads or writes) so didn't justify a conversion of any production volumes from RAID5 to RAID1+0.
(Then again, a 12-disk RAID5 array was no faster than a 6-Disk array for reads or writes either, and a single-pair mirror (RAID1) was as quick as the 6-Disk RAID 1+0 mirrored triplet for all the same tests).

Thanks for the IOmeter info, I might get to try that out, although my main concern now is that a disk failed in the MSA controller cabinet and it somehow caused the connected server to dismount its volumes with device-failure errors rather than continue blithely as should have done.

[CA1004210]

G'day Cameron,

IOMeter is a great tool, indeed. Seeing as how you are at it, you might like to try a few others - I know you can easily get carried away when it comes to throughput testing of any sort....confirming if the first tool gave you the right result or the next one or the next one....

PAT Tool from HP - don't expect Unix standard raw disk device throughput testing with this tool tho - its truely a Windows tool - only checks the mounted filesystem or logical disks ...
http://h20000.www2.hp.com/bizsupport/TechSupport/SoftwareDescription.jsp?locale=en_US〈=English+%28US%29&pnameOID=12188&prodSeriesId=42846&prodTypeId=12169&basePartNum=co557&locBasepartNum=co557en&los=Microsoft+Windows+2000&tech=Software

HP Read Data Tool:
http://h20000.www2.hp.com/bizsupport/TechSupport/SoftwareDescription.jsp?locale=en_US〈=English+%28US%29&pnameOID=12188&prodSeriesId=42846&prodTypeId=12169&basePartNum=COL3285&locBasepartNum=co-11016-1&os=Microsoft+Windows+2000&tech=Utility

HP Create Data Tool:
http://h20000.www2.hp.com/bizsupport/TechSupport/SoftwareDescription.jsp?locale=en_US〈=English+%28US%29&pnameOID=238682&basePartNum=COL1945&locBasepartNum=co-9515-2&os=Microsoft+Windows+2000+Pro&tech=Utility

Nbench - nice and quick - GUI is not flash!!
http://beta.acnc.com/04_02.html
they have a few other disk benchmarking tools on this site - you might like to give them a try too....

Good luck.

Regards,
Saket.