Now, to see how this thing competes. Iomega specifications indicate that the Rev 70GB system offers transfer rates of 30MB/sec. Of course, we're testing the drive when housed in an external USB enclosure, rather than an internal IDE model (which is also currently available; a SATA model of the REV drive should be on the market within weeks). On paper, USB 2.0 supports transfer rates of 480 megabits/sec (that's 60 MB/sec). In reality, we all know the transfer rate is somewhat less -- and that's not taking into account the innate read/write speed of the media, plus OS overhead.

We'll start with straight-up synthetic benchmarking. We chose PerformanceTest 6.0 for this task. It's not as well-known as some products like SiSoft Sandra, but one chief thing it has going for it is support for devices like the Rev -- whereas Sandra and some other higher-end benchmarking software can't quite figure out whether the Rev is a CD/DVD or a hard drive, and doesn't adequately test it either way.

First, a look at our test setup. We're going for very average specs here, since the Rev 70GB could ideally be in use in systems ranging from low-end PCs in the classroom, to mid-range office PCs, to higher-end server setups.

Initially, we ran a simple Read test in PerformanceTest 6.0 (using the standard Win32 API, no caching, 32k block size) to see what sort of numbers the drive generated from reading data across the media.

Read Speed: 10.8MB/sec, roughly equivalent to 8x DVD read speeds.

Let's get a bit more in-depth. We then undertook two sets of tests using PerformanceTest, both using the standard Win32 API, with no caching, and 32k block size.

First, we'll look at how the two drives compared in head-to-head sequential read and write tests. By “sequential,” of course, we mean tests in which data is written or read unbroken across an area of the media. This is a meaningful test in many type of scenarios -- backups, among them.

Judging from the results, the Rev's sequential write and read speeds put the Vantec-Maxtor setup to shame.

The opposite of a sequential data test is one in which reads and writes are randomized across the disk. Let's take a look at how the Rev fares in tests in which 50% of reads and writes are randomized. This more closely emulates normal hard-drive activity, which is rarely limited to one contiguous stretch of media.

As we can see, the Rev performs rather poorly when it comes to non-sequential access. In other words, moving from one area of the Rev disk platter to another is a relatively time-consuming action.

From these tests, we can conclude that the Rev is far more suited to long periods of sequential disk activity -- again, like backups -- than for substituting for hard drive activity. Indeed, the Rev seems to be ideally suited for just such a task, like DAT tapes, but with the added benefit of random, if slow, access.

That isn't to say you can't use the Rev as, say, your main hard drive, or your Photoshop CS2 scratch disk -- it merely wouldn't be very performance-minded. And we wouldn't recommend building an array of Rev disks as a RAID 0 array, either -- unless you're feeling masochistic.

Next, we decided to see how it fared using its own backup software, EMC Retrospect Express. We pitted the device against an inexpensive Maxtor 160GB drive housed in a Vantec external/network drive enclosure (which we reviewed here.) We backed up a 20.2GB partition to both USB 2.0 devices.

The Rev 70GB outperformed the older Maxtor hard drive even when housed in a fairly robust enclosure. Again, sequential write speeds make the difference here.

Finally, a Windows File Copy benchmark. We ran the copy test consecutively over the course of two days, using a ZIP file of 981,644 KB in size, and took the arithmetic mean.

Once more, the edge in a sequential write test goes to the Iomega Rev 70GB. Reading from the devices and writing to a hard drive, however, shows the Vantec/Maxtor setup faring only slightly better, an indication of the significant slowdown the Rev experiences in a partially-randomized data-reading scenario.