|
Figure 7 – Asynchronous throughput using no file system buffering: NT4SP6 vs. NT4SP3
|
| bet | 16/47 | | Sana | 30.03.2021 | | Hajmi | 1.33 Mb. | | #13789 |
Figure 7 – Asynchronous throughput using no file system buffering: NT4SP6 vs. NT4SP3. Read and WCE write throughput for small requests has improved in SP6 as 2KB and 4KB requests now generate the same throughput as larger request sizes. The right graphs show a higher peak throughput as they were using faster SCSI drives. Unbuffered writes were measured on a ST34572 drive instead of our ST34371. The lower unbuffered write throughput we were able to achieve on NT4SP6 is likely due to differences in the drives under test, rather than due to differences between NT4SP3 and NT4SP6.
|
Depths beyond four showed no additional throughput gains. Writes of three deep and beyond reach the same performance as WCE at 32KB requests.
|
| |
