[Posted Monday, March 12th]
Last month, I offered a "first look" at Apple's new AirPort Extreme Base Station. It was written after only a couple of days experience with the new device. Such articles have their value, especially for those who are considering an immediate purchase and want the instant feedback that has become the modus operandi on the Web.
Still, there is remains a useful place for less hurried more deliberate analyses. With this in mind, this month's column takes a second look the new AirPort Base Station, providing results from more extensive testing, several tidbits of information from Apple, and a surprising (at least to me!) conclusion.
Bandwidth and speed
The central feature of the new Base Station is its support for 802.11n networks, claimed (by Apple) to be as much as five times faster than the 802.11g protocol used in the previous Base Station model. In last month's column, I compared download speeds from a new Base Station for different network configurations. This is not a good test for assessing the potential speed benefits of a "n" network, as the maximum download speed from a typical cable modem is slower than the typical bandwidth of a n network. Still, I expected that download speeds on an n network would at least be equal to that of other network types. Unexpectedly, that was not exactly what I found last month. This led me to do additional testing for this month's article.
Download tests. For comparison purposes, I repeated the download measure I ran last month, using the pcpitstop.com download bandwidth test. I also added a new download test: the time needed to download a 5.5 MB file from Apple's Web site. Using a Core 2 Duo MacBook Pro connected to a new AirPort Extreme Base Station, I tested 3 different basic network settings (as set from the "Radio Mode" pop-up menu in AirPort Utility): n-only at 5 GHz (with wide channels enabled), n-only at 2.4 GHz, and n b/g-compatible at 2.4 GHz. I tested each setting two different ways, once with a WPA password enabled and again with no password enabled (as set via the "Wireless Security" pop-up menu in AirPort Utility). This led to a total of 6 different test setups. In all cases, I used the "Share a public IP address" option for the Internet connection (as set via the "Connection Sharing" pop-up menu in AirPort Utility) and the MacBook Pro was the only device on the network.
For all test conditions, download speed was significantly faster when WPA encryption was turned off.
For example, looking at the n-only 2.4 GHz setting, the pitstop.com-reported speed was 6409 Kb/s with WPA off vs. only 1333.50 Kb/s with WPA on. For the Apple file download test, speeds were 6.89 seconds with WPA off vs. 9.32 seconds with WPA on. The scores for the b/g/n setup were essentially the same.
Similarly, for n-only at 5 GHz, the pitstop.com difference was 2800 vs. 900 Kb/s and the Apple download comparison was 6.52 vs. 14.26 seconds. While the directional pattern was the same as in the other network setups, the n-only 5 GHz scores were overall slower. In fact, the n-only 5 GHz WPA-enabled scores were the lowest of any scores in the test.
I contacted Apple about this and they confirmed that enabling WPA encryption will exact a performance hit when downloading files, although they were not precise about how much of a hit to expect.
[Note: While I don't recommend going without a password just to get the potential speed boost, it might be an acceptable risk if you enable Access Control to limit which machines can join the network. You could also make it a closed network, so that casual snoopers won't find it.]
Finder copy tests. I attached a USB hard drive to the Extreme Base Station and used the Finder to copy a 102MB file from my MacBook Pro to the hard drive. The effective transfer rate of USB drives (estimated at about 320 Mbps) should exceed the maximum bandwidth of an n network connection (listed at 270-300 Mbps). Thus, this appeared to provide a better assessment of the potential speed benefit of the new AirPort Extreme than the download tests.
The result was that all speed differences among the 6 groups were eradicated. They all took about 35 seconds to transfer the file. Of particular note, unlike the download tests, WPA encryption had no debilitating effect.
I next compared these Finder-copy results to two more conditions: (a) Making a connection to the drive through an older b/g AirPort Extreme (connected via Bridge mode to the new Extreme Base Station ? using an Ethernet cable, a possible constraining factor) and (b) making a copy to the USB drive when it was directly connected to a USB port on the MacBook Pro. In the former case, the speed slowed down to about 54 seconds. In the latter case, the transfer time zipped along at less than 7 seconds!
Figure 1. Finder transfer speeds from a MacBook Pro to a USB hard drive (see text for more details)
Multiple devices on the network. To assess the potentially-slowing effect of adding 802.11b/g devices to a network, I repeated the n b/g-compatible network tests with my MacBook Pro but now with two 802.11g devices (a Power Mac G5 and an Intel iMac) attached to the network. The results? Adding the two 802.11g devices had absolutely no effect. The speeds across all tests were the same (or even slightly faster) than when the MacBook Pro was the only device on the network.
To be clear, the additional devices were sitting idle while I was testing the MacBook Pro. If the added devices were active (such as all downloading files from the Web) during the testing, I would have expected to see an overall slowing down. But this would likely be true for any network setup. My point here is that simply adding a b/g device to a n-capable network did not by itself lead to a slow down for the n-connected device. However, I have talked to Apple about this issue. They contend that simply adding a b/g device to a n network will inevitably slow the network down, with all devices communicating no faster than the slowest device on the network. While my limited tests did not find this to be the case, I trust Apple as being correct. That's why I have set up my wireless system so that only n-capable devices are on the n network (as described in last month's column).
Rates reported in AirPort Utility. Launch AirPort Utility, open the configuration settings for a Base Station and click on the Advanced icon in the toolbar. Next, click the "Logs and Statistics" button at the bottom of the window. From the new screen that appears, click the "Wireless Clients" tab. This leads to a display of a graph and table indicating the performance of each wireless client connected to the Base Station. Of most interest here is the Rate column (which appears to be listing the speed of the connection in Mbps).
I initially used these tables to compare the rates of Macs that were located within a few feet of my Base Stations. For starters, devices connected to my older b/g AirPort Extreme checked in at the maximum rate of 54 Mbps. For the new 802.11n AirPort Extreme, I compared results across the same 6 conditions previously described. The rate was a rock-steady 130 Mbps in all but the n-only 5 GHz condition. This is approximately what the "real-world" expectation is for n networks.
For the n-only 5 GHz condition, the listed rate jumped to as high as 300 (ranging over time from between 243 - 300). In other words, AirPort Utility indicated that the n-only 5 GHz condition should be the best performer. My test results did not support this. It was no faster than the other conditions for the Finder copy test. For the download test (although, as I indicated, not appropriate for assessing a speed increase in a n-network), it was actually slower than the other connections.
Figure 2. AirPort Utility's Wireless Clients screen (shown for an n-only 2.4 GHz connection)
Signal strength, range, and rates
Apple advertises the new AirPort Extreme as not only faster than the previous model, but having "up to twice the range of networks created with the earlier 802.11g standard." So I decided to run an informal test of this claim. I moved my MacBook Pro to my living room (near where I will be connecting my Apple TV when it arrives). This is only about 30-35 feet from the Base Station, on the same floor. From here, I checked: (a) signal strength (as indicated by the number of bars in Internet Connect) and (b) bandwidth rate (as indicated by the rate in the Wireless Clients display of AirPort Utility). What I found was, once again, a bit surprising.
In the n-only 2.4 GHz and n b/g-compatible conditions, the rate hovered at around 78, dropping down from the 130 rate achieved when in my office. The signal strength bars remained hovering at or near the maximum of 15, the same as when the Mac was in my office.
However, when I switched to the n-only 5 GHz setup, the rate dropped down to between 6 and 40 (usually closer to 6), while signal strength dropped down to around 4-6 bars. Occasionally, the signal became so low that the network name disappeared entirely from the AirPort menu and the connection was lost. This never happened in any of the other conditions. Even my older AirPort Extreme did better, maintaining its 54 Mbps rate with 15 bars! Whether or not WPA encryption was enabled had no effect on any measure here.
As this drop-off seemed contrary to Apple's claims, I again contacted Apple for an explanation. They replied that n networks do have a potentially wider range but that ? for 5 GHz networks ? the range may be "dramatically reduced by solid mass penetration." In plain English, a "solid mass penetration" means a wall. A Wikipedia article, citing data from a 2007 IEEE draft, lists the expected range as the same (~ 50 meters indoors; ~125 meters outdoors) at either GHz but makes no mention of the wall issue. Assuming Apple is correct, this means that the range in 5 GHz mode, when going through a wall, can be less than 12 meters (as was the case in my test)!
Two minor notes
The problem I reported in last month's column, regarding password errors, turned out to be due to an error in the settings of my two AirPort Express Base Stations. When all my Base Stations were configured correctly, the password error messages vanished.
Reviews of the new Base Station at both MacInTouch and Macworld describe a bug, confirmed by Apple, where a Base Station with NAT enabled and connected to a larger network over the WAN port will have slower throughput to the larger network than to devices connected wirelessly or via a LAN.
For a typical user who employs a Base Station primarily as a means to wirelessly connect to the Internet and occasionally transfer files between Macs, the advantages of the new Base Station are not enough to justify replacing an older Extreme Base Station. There are several reasons for this (none of which are clearly stated in Apple's advertising for the new device):
- Your Internet connection speed is not likely to be any faster than with a previous generation Base Station, as typical broadband modem speeds are not fast enough to take advantage of the potential increased speed of a 802.11n network.
- In fact, as confirmed by Apple, with WPA encryption enabled on the new Base Station and especially if you also use the n-only 5 GHz network setting, there is a reasonable chance your Internet connection speed will be slower than with the older Base Station.
- If you use the new Base Station's 5 GHz setting (desirable to avoid interference with other nearby 2.4 GHz devices), and there is a wall between the Base Station and your computer (or other wireless client), your reception range may be so poor as to be unusable. Again, this was confirmed by Apple. [The range limits of the new Base Station's 2.4 GHz settings, or the 5 GHz setting without wall interference, may exceed those of the older model; I did not assess this.]
- If you need to use the new Base Station's n b/g-compatible mode, presumably because you intend to mix n and g clients on the same network, Apple claims that n-capable clients will lose much, if not all, of the speed advantage of the n network. My minimal testing (of Finder copies to a USB hard drive connected to the Base Station) did not support this conclusion; but perhaps more extensive testing would have done so.
The only measure I assessed where the new Base Station clearly surpassed the older one was when doing a Finder copy to a USB drive, when both the Base Station and the client were in the same room. More generally, across all tests, the best results I achieved were with the two 2.4 GHz modes (n-only and n b/g-compatible), with encryption off.
For users considering getting their first AirPort Extreme, the new model is an attractive product if still a bit overpriced compared to third-party alternatives.
Update on creating custom bootable discs and drives. Recently, I wrote a MacFixIt column describing how to make your own bootable optical disc or flash drive, customized with the utilities of your choice. Interestingly, two vendors have now released software that accomplishes the same task. Perhaps this is a timely coincidence or perhaps the vendors were spurred by my column into thinking there was software niche worth filling here. In either case, your choices for creating a bootable flash drive now include Micromat's $135 ProToGo and SubRosaSoft's freeware DasBoot. The Micromat utility includes TechTool Pro 4 and DiskStudio. DasBoot comes with no additional software; plus it requires that you own a commercial bootable CD (such as DiskWarrior or TechTool Pro) to use as a starting point for building your system. Alternatively, you can build your own bootable flash drive entirely for free. Expanding on the instructions in my MacFixIt column, I provide step-by-step details in this article: Creating a custom bootable 1GB flash drive. If you are willing to pay for a shareware program, the article even explains how to add a Finder-like capability to the drive.