How To Create A System Repair Usb

USB 3.0 in the Real World

This site may earn affiliate commissions from the links on this page. Terms of use.

I've written a lot about USB 3.0, and finally I've been able to test an external USB hard drive in the real world. As I expected, I did see a big improvement in performance, but not nearly the performance numbers I saw in the demos at CES. In part, that's because the speed of the USB connection is only one of many factors that impact real-world hard-drive performance.

The drive I tested was the Seagate Black Armor PS110, a 500GB, 7,200-rpm, 2.5-inch external drive. It comes with an Express Card USB 3.0 controller (based on NEC's internal controller). With a list price of $179.99, and features such as 256-bit AES encryption, this drive is aimed at enthusiasts and a business audience, since it's designed to be faster than the company's more consumer-oriented 5,400-rpm FreeAgent drives.

The Black Armor drive is designed for backing up laptops, with features including backup software and encryption. It can be attached to a USB 2.0 port or a USB 3.0 one (which looks the same but is electrically different, though backwards compatible). Because almost no machines come with USB 3.0 ports yet, the drive ships with an ExpressCard USB 3.0 connector and software that makes it work with a Windows laptop. Because ExpressCard doesn't carry as much power as USB, you also have to plug the card into a USB 2.0 port.

To test performance, I copied a subset of my photos, directories containing 26.7GB of data (16,841 items), to the drive when it was connected to a normal USB 2.0 port and to the USB 3.0 controller; and I also read the same data back to another directory on the host computer. For comparison, I tried the same tests with a 500GB FreeAgent drive. And because transfer speed depends in part on the speed of the host computer, I tried it on two different laptops: a Lenovo Thinkpad T400s with an SSD and a Hewlett-Packard Pavilion dv6 with a hard drive.

Here are the results I got. (Note that "read speeds" means reading from the external drive and writing to the internal one, and "write speeds" are the reverse.)

USB 3.0 in the Real World - Test Results

A few things stand out here. To begin with, none of these numbers is anywhere close to the numbers I was seeing at demos at CES, which boasted of 100-MBps read times and 67-MBps write times. I didn't expect transfers at anywhere near the maximum rated speed of the connections (USB 2.0 theoretically gets up to 480 Mbps or 60 MBps, and USB 3.0 up to 5 Gbps, which would be something like 600 MBps). The protocols always have lots of overhead; and on USB 3.0, the performance will undoubtedly be limited by the physical drive, how fast it spins, and how quickly the heads move within the drive. But I was expecting much more of an improvement.

So what's going on? In part, the demos tend to be of sequential reads and writes, but in the real world, most people backing up data end up doing random reads and writes with lots of relatively small files. In addition, the number of files turns out to be significant, and Windows not only copies the data but also writes the location to the file allocation table within its file system (NTFS). Writing lots of small files takes notably longer than writing a few large ones. In addition, adding more cache on the hard drive itself should impact the times. And of course, there's also overhead added by running tests on a real system as opposed to one specially tuned to get the best possible results.

There were other things I didn't count on. In general, I did see slightly better performance on the 7,200-rpm drive than on the 5,400-rpm one, which was expected. But sometimes I saw a big advantage when connected to a machine with an SSD as opposed to a hard drive, and sometimes I didn't. In fact, I saw a lot more variability in what I was thinking of as read speeds on the SSD machine (in other words, when it was reading from the external drive and writing to the SSD). My guess is that happened because the SSD had seen a fair amount of use in the past couple of months, and that sometimes it had to slow down to erase certain sectors (something that SSDs need to do separately, but hard drives don't). I've listed the fastest time.

In addition, I should point out that in all my tests, I did see some variability (about 10 to 15 percent) in different runs of these tests. That's probably because it matters where on the disk the data is written to, so don't worry about any minor differences in the numbers.

So what's the bottom line? In all cases, I did see a notable performance improvement using USB 3.0, but it wasn't anywhere near the 10 times improvement in rated connection speed, or the two-to-three times improvement I was hoping to see. Still, writing at 24 MBps is a lot better than at 14 MBps, and the difference in price is fairly small, so I can recommend these drives as a real improvement. I just wanted more.

For more on Michael Miller's take on technology, read his blog Forward Thinking.