In a previous entry I talked about this new gadget I had ordered, the Netgear SC101, and how I was growing skeptical of the wisdom of my purchase even before I had taken it out of the box. And when I said that Netgear was lying in the product’s package, I wasn’t kidding — the top of the box promises that the SC101 can use “any hard disk from any vendor”, when this is not only not true, but Netgear is aware that it is not true.
Be that as it may, you obviously can’t have a review without actually putting a product through some use, and so I did go ahead and do just that. And as it were the product itself is not all bad, but it’s not really good. Still, let’s start on a positive note.
Product Design
This is a really nicely designed product. As you can see by the various product pictures available on the internet, it’s quite a cute little device. In fact as the box was sitting on my desk lots of people remarked that it looked like a little toaster, and it does indeed resemble a scaled-down kitchen toaster. Its lines are very clean and I’d even describe it as stylish in a Braun appliance kind of way.
It’s also a good functional design, in that setting up the product is a foolproof process. It’s not very often that one can say that about a consumer-targeted computer peripheral. One literally can not set up the product incorrectly. Installation requires no fiddling with screws, the cables are just long enough to restrict the user to the proper installation position, and the only thing you need to get started is a coin that lets you open the device’s front panel. On the design front, it’s a pretty resounding success.
Technologically the designers had a pretty good idea for handling heat too, although the execution might be questionable. The SC101 uses passive cooling; there are no fans. Essentially the whole device acts like the sort of heatsink that you usually see put on a CPU; the inside of the SC101 is composed of a block of aluminum which basically takes the heat from the hard disks and dissipates it through little fins at the top of the aluminum blocks. In theory at least this should work perfectly.
There shouldn’t be, but there is… CPUs don’t have moving parts, while hard drives have lots of them. They spin up and spin down; they vibrate when this happens, and this affects the drive heads; if you installed a hard disk to stay as tightly as a CPU it would break down very quickly. So while the heatsink idea is nifty, it’s ill-suited to hard disks, because you just can’t have the sort of tight metal-to-metal contact that makes a heatsink work so well when it’s used to cool a CPU. Clearly the SC101’s aluminum chassis absorbs and dissipates quite a bit of heat — enough that Netgear deemed it necessary to warn you that the SC101’s “surface may get hot” — but heat IS a concern when the drive is in heavy use, particularly when it’s configured as a mirrored drive.
Testing
I wasn’t exactly sure how to test this drive, or more exactly what to compare it to when it came to measuring performance. The fact is that this is a two-disk PATA device, and in that sense it’s fairly unusual; usually multiple-disk device will use either SCSI or SATA drives. Nothing very specialized. Basically I just did a file copy from my PC to the device in a way that would let me measure the throughput achieved without extraneous factors; so the transfer size would have to be large enough to be significant, it had to be done in large chunks, and it would have to take place in something resembling real-world situations.
For me — and it may well not be the case for others, but I can only speak for myself — good write performance on large files is essential. I bought this so I could have a sort of vault for the digital photos I take. I have a Canon Digital Rebel XTI, so I’ll be working with Camera RAW files that are roughly 8 to 10 megabytes in size. I will be transferring blocks of photos roughly 4 gigabytes in size from the camera to the storage device. And I’ll probably have torrents open and want to do some light web-surfing while this is happening. Nothing big or too processor-intensive.
So I needed a multi-gigabyte file transfer from the computer to the device while still using the PC for some light tasks. I grabbed a 35GB folder containing some 155 equal-sized OGG media files, and dropped it on the SC101’s disk icon. The drive was set up in a mirrored configuration, which Netgear encourages, so in my mind this is probably a typical use for the SC101. Well, maybe 35G is not “typical”, but the thing should be able to accomodate large file transfers. Using that particular folder also made sure that the overhead of creating new files, etc. wouldn’t be too much of a burden on the device and skew the performance.
Frankly, the SC101 didn’t do too well at this. You know how people sometimes joke that they can start a program, go and make a cup of tea/coffee, and come back before the program is started? Well with the SC101 you can run the test I describe, go out to dinner 100km away, come back, and your task will still be running. It took a whopping 3 hours 37 minutes for the copy to get completed. That’s well over 6 minutes per gigabyte, 161MB/minute, 372K/second.
These are not good numbers. I can write to CD or DVD faster than that. And I’ll be able to read those CDs and DVDs on a Linux box, which isn’t something I can do with the SC101.
Numbers don’t exist in a vacuum, though. To be fair I must have something to compare this device with that’s in the same class and will let me do more or less then same thing. To make it a true comparison I must also get a device that is able to take the very same two drives I used in the SC101, set them up in the same way (mirrored), and redo the same test with the same parameters. Only then will I be able to tell how the device’s performance really stands.
So I went on the net and ordered a PATA-RAID card from 3ware, the Escalade 7006-2. It has two parallel-ATA ports (same as the SC101), so the exact same drives will be used, and it only costs slightly more ($135 vs. $101). Installation isn’t quite as straightforward, but then I don’t think performance will be as anaemic, either.
Unfortunately I can’t yet tell the results of the test because I haven’t yet received the device. So this article series is not over.