It's not so hard to measure this yourself. Just try several configurations
and any disk benchmark software will show throughput increases when
adding more spindles.
At a certain point, you will hit other bottlenecks, and adding more
disks won't
help. In older systems, the parallel SCSI bus often meant using more than
four disks for a single array was useless. On a SAN system I recently
configured,
20 SATA disks was the max. With SSD's on a LSI Raid controller, I
measered 8 disks
to be the max. I would think more than 8 SAS 15kRPM disks in a single array
will not help either.
Splitting disks across multiple arrays will decrease your access times,
which is
very important for databases. So ideally you would have multiple
RAID-10 arrays
to have both good throughput as well as fast access times.
Pim
and any disk benchmark software will show throughput increases when
adding more spindles.
At a certain point, you will hit other bottlenecks, and adding more
disks won't
help. In older systems, the parallel SCSI bus often meant using more than
four disks for a single array was useless. On a SAN system I recently
configured,
20 SATA disks was the max. With SSD's on a LSI Raid controller, I
measered 8 disks
to be the max. I would think more than 8 SAS 15kRPM disks in a single array
will not help either.
Splitting disks across multiple arrays will decrease your access times,
which is
very important for databases. So ideally you would have multiple
RAID-10 arrays
to have both good throughput as well as fast access times.
Pim
On 21-2-2011 7:40, randyduly wrote:
> Does anybody have any documentation either from Epicor or else where to prove why RAID 10 with more disk platters is better for Vantage 8 or Epicor 9. I am trying to make an arguement for getting a different server with RAID-10 and more disks.
>