Seagate ST3320620A Barracuda 7200.10 (3.5″ 320GB 2007)

The chief competitor to the WD Caviar SE was the Barracuda 7200.10 in a market which was increasingly becoming a “two-horse race”. Unfortunately, as we reached the limits of what conventional longitudinal magnetic recording could offer, reliability was not good with this 320GB unit either.

The Drive

A mostly-flat top lid is carried forward from the design seen in the 7200.7 generation but with a minor change to add an extra screw in the top-center.

On the rear, everything has mostly been integrated, thus the PCB has been further reduced in size. Previous initiatives to provide protection in the form of a shield or cover have not returned, seemingly being completely abandoned.

Labelling of the serial number is on the front edge, while the side rails continue to have milled channels in them (probably for manufacturing use), but in this case, do not completely go through the aluminium as with previous generation drives.

Performance Testing

CrystalDiskInfo

This drive did not see much use in my server, as it started to cause problems early on with data loss. It has firmware 3.AAE with a large 16MiB cache buffer, but despite only doing 2144 hours, has 10 reallocations.

Post-testing, this number increased to 24 with a whole host of uncorrectable errors reported to the host. This is what pushing the boundaries (and adding age) seems to do to these drives. That being said, judging from the attributes, this could be one of the earlier drives to implement an active fly height control for the head.

HD Tune

Being a year newer than the Caviar SE 320GB unit I rested, I was pleased to see the drive had improved throughput, reaching an average of 62MB/s. A precipitous dip in the write transfer rate can be seen, along with some “muddiness” in the outer zones due to general drive difficulties – but unlike older drives, it seems this drive uses a very large number of zones to optimise recording density across the surface.

The IOPs results shows no undue hang-ups win reading or writing, although the number of IOPs is necessarily mechanically constrained at the low end of the scale, The effect of the cache is not so apparent, especially with the outer-zone throughput starting to approach interface limits.

CrystalDiskMark

Relatively good performance figures from CDM, which is to be expected, besting the Caviar SE from one year earlier on most metrics.

ATTO

Performance across different access sizes seems to reach near-maximum at 8kB with a relatively even preference between reads and writes.