A few years ago I moved towards eSATA connectivity for drives. Â SATA (or Serial ATA) is the connection that is on the actual drive and has blazing fast transfer speeds up to 3Gb/sec. Â The concept is that whatever drive enclosure you buy has some sort of interface card that ‘converts’ SATA to USB or Firewire. Â That’s the definition of a bandwidth bottleneck. UsingÂ SATAÂ toÂ SATAÂ connectivity should eliminate any bottleneck and give you the full bandwidth of the drive. Â eSATAÂ is simply an external version of theÂ SATAÂ connector. Â I’ll do more detailed explanations in later posts, but in general to useÂ eSATAÂ you need a PCI host adaptor card and a drive with anÂ eSATAÂ connection.
I was excited when the FreeAgent Pro came out because of the eSata connectivity. Â I got a great deal and paid around 100 bucks at Fry’s for a 1TB model. Â Every other article I’ve seen of the FreeAgent Pro drive gives it a rave review and claims that the 3Gb/sec is blazing fast. Â It should be – but it’s not. Â It’s also important to note that these reviewers probably never tested the eSATA connection for speed. They’re just quoting the Seagate data sheet – even using the same word to describe it. Â If they actually tested it or benchmarked it they would report honestly on how the drive is actually performing. No single drive (today) performs at 3Gb/s because that is the maximum burst bandwidth of the drive interface. Â In practice you’re average is going to max out at 1/4 – 1/3 that rate.
Seagate’s datasheet on the (now) FreeAgent Pro Classic has this to say: “ItÂ provides eSATA connectivity at blazing speeds up to 3Gb/sec, FireWireÂ®Â 400 connectivity for Macs and digital video users,Â as well as USB 2.0 connectivity, the most commonly usedÂ interface in the world today.”
The drive seemed fine to me until I tried copying a large chunk of video files to it using the eSATA connection. A copy function that should have taken only an hour or so was claiming that it would take 30 hours! So I decided to start taking some benchmarks and get to the bottom of the problem.
The benchmark tool I’m using is the AJA System Test which is a free app. Â The tool assumes that you are using the drive for video and allows you to choose a frame size and a file size. It then give you the write and read results in Mb/sec and shows a graph which tracks the Mb/sec for each frame. Â This is a good test to see what is happening and I got some very interesting results.
The tests were conducted on my new Mac Pro 8-core. The first test was over USB. First checking system profiler shows that the speed of the drive is “Up to 480 Mb/sec.” Â The small “b” indicates that this is megabits. Â To get megaBytes you divide by 8. Â This means that the data path should have a limit of 60MB/sec. I’ve grabbed tons of screenshots of these tests and will make them available to anyone who asks. Â To keep this post a reasonable length I’m mostly only posting the images of the graphs.
The USB test came in at a very consistent Write: 28.1 MB/s and Read: 38.3 MB/s. Â This is about half of the capability of USB 2.0:
Next I mounted the drive using the firewire connection. The FreeAgent Pro has a firewire 400 port. Â The system profiler showed a connection speed of “Up to 400 Mb/sec”. Â That test also showed very consistent numbers. Â Write: 33.5 MB/s and Read: 38.2 MB/s.
These result are fairly typical for USB and Firewire 400 devices.Â But the eSATA should be much faster. Typical numbers there for a single drive should be anywhere from 70MB/s to over 100. Â
Clearly something is not working as advertised. Connecting to via eSATA had very inconsistent results. The data would flow at a good clip and then basically just stall for a few seconds, then jump back up and then stall again. This blew the average bit rates down into the single digits. Â This explains the 30 hour copy. Â Being kind, here is one of the better results (pictured).. Â Write 10.4 MB/s Â Read: 61.5 MB/s. Â But I had seen write results as low as 2 MB/s. Â Check out the graph below:
The blue line is the write speed. Â Each time that blue line dips to zero the device stalled for a few seconds. Â The graph is showing the MB/sec per frame. Â It does not show how long the device stalled. However watching the test I can tell you that it stalls for a while. It is also evident by the low average Write speed.
To test the possibility that I just have a bad drive I tried a different FreeAgent Pro. Â That one had very similar results. Â Every few “frames” of data the drive completely stalled for a few seconds and then had a burst of bandwidth. The graph below represents average speeds of Write 5.5 MB/s and Read 47.8 MB/s.
I called Seagate technical support about this and they said that the FreeAgent Pro drive has only been tested on a couple of different cards for the Mac, the SIIG card and the Promise card. Â I tried my drives on two different systems with different eSATA cards. Â The other card showed VERY weird results, possibly because of data caching in the card. Â I have not tried it on either of the two cards that Seagate suggested because I don’t have those cards. However, I have tried other eSATA drives with my cards and they all work as fast as expected (see tomorrow’s review of the Vantac NexStar).
The bottom line isÂ that there should be NOTHING inside the FreeAgent enclosure blocking the data path. Â The SATA on the physical hard drive should be directly passed through to the eSATA connection on the drive. Obviously this is not the case.
One final note is that Seagate now classifies this drive as the FreeAgent Classic and now has newer models. I’d love to test out one of these models but am not willing to shell out the money to see if they’re any better. Â I had high hopes for these drives and I’m glad that the eSATA connectivity is finally getting to commercial drives. Â But this driveÂ disappoints.Â