X-bit labs

New SATA Hard Disk Drives from Western Digital RAID Edition Family (page 4)

Category: Storage

by Andrey Kuznetcov

[ 11/08/2004 | 06:53 PM ]

Performance in Intel IOMeter: WorkStation Pattern

With the help of the WorkStation pattern we imitated the work of the hard disk drive as if it were working in a typical workstation, when the workload is limited by the maximum request queue depth of 32 requests.

In the first diagram you see the situation when the entire storage capacity of the hard disk drives was tested. WD2500PD is generally faster here. The new WD2500SD HDD appears defeated in both cases.

Since the advantage of WD2500PD solution in terms of lazy writing has already revealed itself in DataBase pattern, the results obtained for WorkStation pattern were very easy to predict. Large share of write requests, which is pretty typical of this pattern, doesn’t leave a single chance for WD2500SD: neither in the RAID array, nor in the single-drive mode.

Let’s see how the disks ratings in this pattern compare. To calculate the ratings we apply the following formula, which you should already be familiar with:

Performance = Total I/O (queue=1)/1 + Total I/O (queue=2)/2 + Total I/O (queue=4)/4 + Total I/O (queue=8)/8 + Total I/O (queue=16)/16 + Total I/O (queue=32)/32

The diagram shows that both integral parameters of the overall product efficiency are higher by WD2500PD.

One more situation we decided to check out when running the benchmarks for the WorkStation pattern deals with involving only the first 32GB of storage space of each tested hard disk drive. Again, WD2500PD looks more convincing here, as it outperforms WD2500SD in the regular mode. However, when we have an array involved, the latter manages to beat WD2500SD only in case of high requests queue depth.

The diagram with overall performance indexes demonstrates pretty unexpected results, as it might seem at first glance. WD2500PD hard disk drive wins only in three cases out of four. It yields to WD2500SD when working in an array, due to higher importance of the results obtained for low queue depths.