Sharing some personal experience from previous benchmarks. Here are some general guidelines for determining the number of hard disk drives required for a given I/O load.
Generally, a single disk drive can do this many "physical" IOPS per disk:
15k rpm: 180-210 IOPS
10k rpm: 130-150 IOPS
7200 rpm: 80-100 IOPS
5400 rpm: 50-80 IOPS
In a mirrored configuration:
Disk IOPS = Read IOPS + (2 * Write IOPS)
In a parity (RAID5) configuration:
Disk IOPS = Read IOPS + (4 * Write IOPS)
Summary of the number of operations per RAID type:
RAID 1 and 1/0 require that two disks to be written for each host initiated write.
Total IO = host reads + 2 x host writes
RAID5 (4+1) requires 4 operations per host write = a RAID5 write requires 2 reads and 2 writes
Total IO = host reads + 4 x host writes
Personal Note on RAID5: as you can see here, RAID5 does need more writes compared to other RAID level. my personal opinion on RAID5 is, when you know your IO pattern from your application you can also go with RAID5 to get efficiency and same time a "good performance" when the majority of IO operations are reads and not writes. a typical scenario for RAID5 would be a.e. "Streaming Services" where mainly 100% large sequential reads does happen.
if you are interested more on characterization of IO pattern, the SQL Team has released an really good Whitepaper on this -> http://download.microsoft.com/download/B/E/1/BE1AABB3-6ED8-4…
Storage Top 10 Best Practices
SQL Server Best Practices Article
NT Server and Disk Subsystem Performance
HYPER-V CLOUD DEPLOYMENT GUIDES