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.
You can click the following link to view the tabled results in IOMeter: Database.
We will build diagrams for request queue depths of 1, 16 and 256.
At the shortest queue depth this test is largely a competition of deferred writing algorithms although the Western Digital and Seagate 7200.4 differ from the others at a very high percentage of reads. The HDD from Western Digital is unrivalled in deferred writing whereas the Seagate 7200.4 is poor despite its 16 megabytes of cache. The 5400rpm Seagate 5400.6 looks much better.
When the queue depth is increased to 16 requests, it is request reordering algorithms that come into play. The Western Digital is still good, but the Seagate Momentus 5400.6 is close behind it. The Seagate Momentus 7200.4 is in the lead at high percentages of reads thanks to its large cache and higher spindle rotation speed.
When the queue is the longest, the HDDs deliver higher performance because NCQ algorithms are complemented by the driver’s request reordering. The Seagate 7200.4 slows down here while the Western Digital takes first place, followed by the Seagate 5400.6.
Summing up this section of our test session, the next diagrams will show you the performance of each HDD at five different queue depths.
The Fujitsu has both NCQ and deferred writing, but the latter is far from effective. As for request reordering, the company should still polish it off: the slumps in the graphs don’t look good at all.
Hitachi revised the firmware for the newer model. It is hard to tell how successful this revision is. The performance hit under mixed load (when the ratio of reads to writes is close to 50/50) is eliminated but the efficiency at long queue depths is now lower. So, the company has still got some work to do here.
Samsung is polishing its firmware off in a steady and wary way. It is good that the newer version has acquired deferred writing and request reordering but why is their efficiency so low? Server loads are still a trial for Samsung’s HDDs.
Seagate’s HDDs behave differently. The 7200rpm model is faster at short queue depths but slower than the 5400rpm model at long queue depths. And while the Seagate 5400.6 has good deferred writing, the Seagate 7200.4 has almost none of it. We must confess Seagate’s firmware never ceases to surprise us.
The Toshiba is similar to the Fujitsu in having average-efficiency deferred writing and far-from-perfect request reordering algorithms.
Western Digital’s HDD is almost ideal. Everything works just as needed in it. This company seems to have developed firmware which is highly adaptable for different HDDs.