Web-Server, File-Server, Workstation Patterns
The controllers are tested under loads typical of servers and workstations.
The names of the patterns are self-explanatory. The request queue is limited to 32 requests in the Workstation pattern. Of course, Web-Server and File-Server are nothing but general names. The former pattern emulates the load of any server that is working with read requests only whereas the latter pattern emulates a server that has to perform a certain percent of writes.
The RAID0 and RAID10 arrays built out of the same number of disks deliver the same performance in this reads-only pattern.
The same goes for the RAID5 and RAID6 arrays: they are almost equals. The graphs of the degraded arrays nearly coincide, too.
The performance ratings suggest that when the load is pure reading, it is the amount of disks in the array that’s important with this RAID controller while the type of the array is insignificant.
As soon as there are write requests to be processed, the RAID0 become faster than the RAID10. The degraded 8-disk RAID10 is better than the 4-disk RAID0, so some of the optimization algorithms are still at work.
In the same way, the RAID6 arrays are inferior to the RAID5 ones under this load because they have to process twice the amount of checksums. There is a difference between the degraded arrays, too. Restoring data from two checksums to do a write is not “free” anymore.
Take note that the checksum-based arrays are worse than the RAID10, let alone the RAID0.
It is like in the previous test, but the noted peculiarities are more conspicuous because the Workstation pattern contains a larger share of writes.
Now the 8-disk RAID5 and RAID6 arrays can only compete with the 4-disk RAID0 while their 4-disk versions are hardly different from the single HDD. The degraded arrays are downright poor in this test.
It’s roughly the same when the test zone is limited to 32GB. The multi-disk arrays speed up because the working zone of each HDD is very narrow and the HDD’s heads travel by very short distances.