In the Database pattern the drive is processing a stream of requests to read and write 8KB random-address data blocks. The ratio of read to write requests is changing from 0% to 100% with a step of 10% throughout the test while the request queue depth varies from 1 to 256.
The odd results of the previous tests made us use this test although we don’t usually employ it for testing external HDDs. It can show a drive’s behavior under load very clearly.
You can click the following link to view the tabled results.
Let’s view diagrams with each drive’s graphs at five different request queue depths.
We don’t see anything exceptional with the FreeAgent Desktop. It has average performance, modest deferred writing, and zigzagging graphs. The latter fact indicates imperfections in the drive’s or controller’s firmware, or both.
The newer FreeAgent Desk behaves in the same way. The good news is that this model delivers higher performance in the most complex range of loads, i.e. when there is about the same share of writes and reads. The overall shape of the graphs proves that these devices have identical HDDs inside.
Well, the FreeAgent XTreme has problems with writing indeed. The awful fluctuations of speed at high percentages of writes mean that this drive’s deferred writing algorithms are poor.
And we see the same problems with the two other interfaces. This drive must be unlucky with the firmware of the controller installed in it: deferred writing works at certain combinations of loads only. That’s a depressing performance really.