Aiport Extreme 802.11n

= Apple's Airdisk feature of Airport Extreme is a dud. =

What follows is proof of my assertion. The new feature that is supposed to set this new AE apart from the older base stations is the ability to share a USB attached hard drive to your home network. With 10.5 on the way, this could be a really nice way to implement a Time Machine backup drive for every Mac in your home.

It is this feature that presumably makes the new Airport Extreme worth more than its predecessor, and helps justify being nearly twice the cost of other good wireless access points (like the Linksys WRT54G). It was exactly this feature that I purchased it for.

These tests began this morning (Feb 19, 2007) when I attempted to copy 2GB of files (my ~/Documents folder) to a 300GB backup disk, attached via USB 2.0, to my new Airport Extreme (802.11n) base station. When the copy began to take longer than a few minutes, I presumed something must be wrong and began doing some tests.

Test Methodology
Test results are reported in minutes. The Elapsed and Mb/sec calculations were derived from one-second precision measurements.

The disk image is a 2GB Parallels virtual disk of FreeBSD. The documents is my ~/Documents folder. The similarity in scale makes for a very convenient comparison. The actual size of the disk image is 2,121,706 KB and the size of the documents folder is 2,127,921 KB.

All devices connected to the gigabit network are synced at full duplex. All Mac computers in these tests are running 10.4.8. See the network diagram for further details of network setup. The ezRAID is a two disk external drive chassis with two FireWire 800 ports and one USB 2.0 port. The disks are hot swappable and each disk tray has a built in cooling fan. The disks are rated for 24x7 operation so I can leave them powered up  at all times.

Test Notes
Test #1 & #2 ezRAID connected to AE BS via USB cable. Files are being copied from my 24" iMac so the path through all the devices looks like this: iMac 24" -> GB Switch -> GB switch -> Airport -> USB cable -> ezRAID. We'll test my Gigabit backbone a little later so don't worry about that. What is surprising about these numbers is their abysmal speed. Since the Airport is connected to my LAN via the WAN port, I also repeated this test using one of the LAN ports with similar results.

Test #3 & #4 To see if the problems were related to the USB connection, I connected the ezRAID directly to the MacBook. This test assures that our USB cable is sufficient for USB 2.0, and also tests the ezRAID USB interface. As we can see from the results, USB 2.0 performance is significantly higher using the MacBook USB port instead of the Airport. The location of the bottleneck is quite obviously in the Airport.

Test #5 & #6 These two tests are simply for comparison, so we can see how Apple's USB 2.0 implementation fares against FireWire 400. As the results show, FireWire is nearly twice as fast. The MacBook does not have FireWire 800 ports.

Test #7 & #8 More reference benchmarks. Here we copy the same files from the iMac HD to the MacBook HD. This reference point demonstrates that my Gigabit networks is not a bottleneck. We are able to copy the files to the MacBook and are limited by the speed of the MacBook hard disk.

Test #9 & #10 This is the kind of performance I was hoping for from the Airport Extreme. The USB 2.0 bus speed is rated at 480Mb/s and it seems fairly reasonable to expect that on a device that support disk sharing, it would have a reasonable speed USB implementation. Unfortunately, it is not so. In order to put my drive chassis in the closet and still achieve reasonable performance, it is obvious that my disks must be attached to a computer with firewire, or a much better NAS type device than this new Airport.

Test #11 & #12 The ezRAID directly connected to my iMac via Firewire 800.

Test #13 iMac to PowerBook G4 via Gigabit. Compare to MacBook test #7. The drive in the PowerBook is obviously much slower.

Test #14, #15, #16 Test the performance of the PowerBook's FireWire implementation. In tests #14 and #15 we see the limits of our internal hard disk in the PowerBook. The test results jump back up when we grab the files via Gigabit and write them to disk.