Disk Response Time and Average Positioning Speed
For 10 minutes IOMeter is sending a stream of requests to read and write 512-byte data blocks with a request queue of 1. The total of requests processed by each HDD is much larger than its cache, so we get a sustained response time that doesn’t depend on the HDD’s buffer size.
It is simple with the read response time parameter: the Caviar Black series takes all the places on the podium. The leaders differ between each other, though. The newer E8 is somewhat better than the J7 and takes first place. The new dual-platter models are somewhat worse, the E3’s response being nearly 1 millisecond higher. Interestingly, the dual-platter FAEX is a quarter of a second faster than the similarly designed E3 and seems to really have new read/write heads. We will be sure about that after we test the drive in small traces and using a narrow test zone where the auxiliary piezoelectric actuator is going to deliver higher performance benefits.
Seagate’s Constellation NS is good. This company’s products have not shown such a low response time for long. The Barracuda XT is, on the contrary, poor, being even inferior to the two power-efficient models.
The WD Caviar Green is a disappointment. Its response of 20 milliseconds is too much even for a drive with a reduced spindle speed. We can’t really think of an explanation as its predecessor with ordinary sectors and its 1-terabyte cousin looked better in this test. Perhaps we’ve got a defective sample or a drive from some special series optimized for quiet operation (by lowering the speed of the heads). Whatever the reason, this model will be a poor choice as a system disk or a disk to store a database on. It can do nicely as a passive storage device which only has to deliver a normal sequential speed and consume as little power as possible.
We also have interesting results at writing. WD’s dual-platter Caviar Black have a higher response here, indicating a reduced efficiency of their deferred writing. And comparing Seagate’s HDDs, we can see that the Barracuda LP uses deferred writing very sparingly. The WD Caviar Green behaves in a stable manner, delivering a response time of 20 milliseconds again. This result can be explained easily, though. Our test load operates with 512-byte sectors and this HDD, due to the emulation, has to first read a 4KB data block, modify 512 bytes in it, and then perform a write operation for each write request! The saddest thing is that operating systems perform most of their service operations in such small requests to minimize the risk of losing important data.
Next goes the test of average positioning speed. The drive is being bombarded with read requests like in the response time test, and we calculate the difference between the LBA addresses of the previous and next requests and divide it by the time it took to perform the request. In other words, we have the distance (in gigabytes) the drive’s heads can run through in 1 second.
Besides response time proper, the number of platters is important for this test. The more platters the drive has, the longer distance its heads can run. As you can see, the four-platter 2-terabyte HDDs occupy the podium, ordered according to the speed of their heads. Next go the old 3-platter Caviar Black. Having more platters, they are equal to (or even a little faster than) their higher-density cousins of the newer generation. The Hitachi is poor in this test as it has few platters as well as sluggish heads.