RAID consideration for 24 Disk Array

hardwareraidstorage

I have here a 24 Disk Array with 3.7 TB disks in it. What would be, performance wise a good configuration when using a RAID 6: RAID 6 over all 24 disks or should I use 2×12 disk RAID 6 and than a RAID 0?

I'm not so interested in conversation about the RAID level itself (e.g. 5 or 6 or 10) but more about
the arrangement of the disks. If it would be better using multiple smaller RAID groups or one big RAID group for example… what's here the best practice?

Best.

Best Answer

Some of this hinges on the hardware involved. I prefer RAID 1+0 for simplicity and rebuild times. It's tough to give a generic answer without more details, though...

Things to consider:

The disks installed in the system: SAS? SATA? Nearline SAS? This impacts the failure rate and failure mode, as well as array rebuild times.
The anticipated use for storage: Your performance requirements may drive the design. Random I/O? Sequential? Read-biased? Write-biased?
Interconnects: How will the storage array be connected to the server? SAS? Will you be using a single connection to an HBA? Two? Multipath? 3Gbps? 6Gbps? There will be a ceiling in storage throughput because of SAS oversubscription. So this factors into the design because of that performance cap.
Controller: I always come from an HP SmartArray perspective, but I suppose the rest of the world uses LSI and PERC controllers. This may be a moot discussion, as LSI controllers can't have more than 16 disks in a single-level virtual drive; e.g. you wouldn't be able to create a 24-disk RAID6 volume. You can do this with HP controllers, though.
Resiliency: Do you plan to have online spares? When you consider a nested RAID level like 60, that becomes important.

So, assuming a controller capable of both. Your options are really 4 x 6-disk RAID6+0, 3 x 8-disk RAID6+0, 2 x 12-disk RAID6+0 and a 24-disk RAID6.

Determine the space needs, as they vary. Then evaluate the sequential performance capabilities of each. I'd suggest 3 x 8-disk as a reasonable if you go nested and aren't interested in RAID 1+0.

Related Solutions

Remove 1 Disk From 4 Disk RAID 5 Array

No, and if you have 4 disks and want two spares you are better off performance wise with a 4 disk RAID 10 instead of a 3 disk RAID 5 + hot spare. Either way your array will have 2 disks worth of usable capacity but the RAID 10 will have better write speeds and takes less time to recover from a failed disk.

Assuming the RAID 5 doesn't contain your boot volume, I'd go ahead and backup the data, reconfigure to a RAID 10, and restore the data.

    RAID Level Comparison for arrays up to 8 drives.
Features                RAID 0  RAID 1   RAID 5   RAID 6   RAID 10
Min Drives                2       2        3        4        4
Data Protection           0       1        1        2        2 (Up to one disk failure in each sub-array)
Read Performance         High    High     High     High     High
Write Performance        High   Medium    Low      Low     Medium
Degraded Read Perf       N/A    Medium    Low      Low      High
Degraded Write Perf      N/A     High     Low      Low      High
Capacity Utilization %   100     50      67-87    50-75     50

RAID Array performance on an HP Proliant ML350 G5 Smart Array E200i

Too many imponderables. For example, how are the disks set up? If the logs and data share the same disks the random I/O from the data areas will disrupt the log traffic, which is mostly sequential I/O and is disproportionately affected by a busy random access workload on the same disks.

Without some more insight into your configuration I can't really say what might be causing the problem.

For example, 60MB/sec off a RAID is about right for a 4-disk RAID-5 or RAID-10 with 64k stripes and 15k drives. Each drive will read one 64k stripe per revolution of the disk (about 250/sec for a 15k drive) which gives you 15MB/sec per drive.

The average seek time for a 15k disk is around 3ms across the whole disk. On a mostly contiguous 10GB file on a RAID volume with (say) 146GB or 300GB disks and a bit of help from the cache I could see 30MB/sec being a reasonable figure for a disk array configured as described above. It would indicate averaging a data read about every two revolutions of the disks.

That's a thought off the top of my head for a configuration one might reasonably expect to see on an ML350. However, I have no idea if that matches your actual configuration, so I can't really comment on whether the observations are relevant in your case.

Best Answer

Related Solutions

Remove 1 Disk From 4 Disk RAID 5 Array

RAID Array performance on an HP Proliant ML350 G5 Smart Array E200i

Related Topic