Apple has been playing it cool on the WiFi side of things lately. It started with the previous Airport Extreme (Gen 4) which quietly introduced three spatial stream support, followed up by the Early 2011 MacBook Pro update which brought a three spatial stream compliant WLAN stack, and now has continued with an even more understated update for the Time Capsule (4th generation) and Airport Extreme (5th generation).

Both updates launched just prior to this latest round of Apple launches, which included the Mac Mini, Macbook Air, and Thunderbolt Display, but unlike those three, the Time Capsule and Airport Extreme updates saw almost no mention. Starting with the exterior packaging, you’d be hard pressed to tell that a particular Time Capsule or Airport Extreme is the newer refresh. I no longer have the old Airport Extreme packaging, but the new device box is virtually indistinguishable. Outside of bumping the supported storage capacity for the Time Capsule up to 3TB, there’s no real obvious giveaway for the Time Capsule either.

The only way to tell which version is which by looking at the box is by the model numbers—MD031LL/A for the 5th generation Airport Extreme, and MD032LL/A for the 2TB 4th generation Time Capsule.

The contents of the Airport Extreme box remain the same as well, starting with the device itself on top, and underneath it, a power cable, 12 volt power supply (model A1202) and some literature about setup in a white plastic bag.

The Time Capsule box is much the same affair, with the device inside, a power cable, no power supply (since it’s internal), and some literature.

I stacked all three devices up so you can compare physically. Really the only big giveaway between the two Airport Extremes is an extra line of text on the previous generation, and of course the model number or FCC ID. Both the Time Capsule and Airport Extreme still retain the same port configuration—four GigE ports, one USB 2.0 port, power, reset, and a Kensington security slot. Those four gigabit ethernet ports can either be used as a switch, or you can use the device as a router and then the leftmost port becomes WAN and the right three become LAN.

At this point it isn’t really looking like there’s much different, but exterior appearances can be deceptive.

Apple sent us both a Time Capsule and Airport Extreme which I used for testing, but I also purchased a copy of each to tear down and get to the bottom of the changes internally. Before doing that, however, I hit up the FCC to see what I could glean from comparing the test reports between generations. Apparently I wasn’t alone in doing so, as other people likewise picked up on this avenue for finding out what’s different.

Inside the test reports for both are some nice tables that outline maximum output power for the wireless stack inside the devices. I’ve copied and formatted the data for both the Time Capsule and Airport Extreme.

It’s curious that for the 802.11b category power actually went down on the Airport Extreme, but hopefully nobody will find themselves using 802.11b in the first place. Interestingly enough, the results are almost the same on the Time Capsule, except 802.11b power has gone up accordingly. There are different output powers for each wireless mode, including 20 MHz and 40 MHz channels, but on average power between Airport Extreme generations has increased 135 mW, and 143 mW between Time Capsule generations.

Of course the next logical question is whether antenna gain has changed between the two—after all, having a more powerful output power only goes so far. It turns out that both Airport Extreme generations share the exact same antenna configuration and gain, and likewise with the Time Capsule. Note that two antennas are actually shared among the 2.4GHz and 5GHz RF chains—AP2 and AP3 to be exact.

I hadn’t looked up the Time Capsule antenna gains until now (having not owned one) but it’s surprising how little gain there is on 2.4GHz with antenna 2. I’m a bit surprised that Apple hasn’t moved over to using the antennas from the Airport Extreme in the Time Capsule. The Airport Extreme has a much more even gain configuration between the three antennas, however as we’ll show later performance is relatively comparable between the two products.

Without even breaking the devices open, we now know that the Airport Extreme and Time Capsule both have substantially increased transmit power, and likely share the same radio given the identical transmit power characteristics.

Without taking a peek at the internal FCC photos, I broke out a heat gun and dove inside. I began by opening up my existing Airport Extreme 4th generation. Disassembly still starts by using the heat gun to warm up the adhesive attaching the rubber shoe on the bottom to the metal underside, peeling that off, then removing some screws.

Inside, the Airport Extreme is surprisingly elegant. There’s a single large PCB which serves as a mainboard into which the WLAN card (which is a PCIe x1 slot) plugs.

On that main PCB is a Marvell 88F6281 SoC with an internal ARMv5 Sheeva CPU core clocked at up to 1.5GHz. The Marvell 88F6281 also contains a PCI Express x1 controller, two SATA II ports, a single USB 2.0 port controller, and two GigE ports, not to mention a DDR controller.

Adjacent to the SoC is a Hynix DDR2 SDRAM marked H5PS1G63EFR, which corresponds to a 1 Gb (128 MiB) 400 MHz die.

Switching those ports on the back is the task of a Marvell 88E6350R 7 port GigE switch, with 5 physical interfaces, all of which support up to 10 KByte jumbo frames. Of course, both devices only expose a total of four ports, but in theory there could be five. Those two others are internal for interfacing with the corresponding two ports on the 88F5281 SoC.

There’s also a curious looking button cell 3V battery on the device, which probably keeps the system clock ticking. With no obvious NAND TSOP onboard, I went searching for storage, and found it on an MXIC flash memory marked MX25L12845E which provides 128 megabits (16MB) of flash storage.

But probably the most interesting component on the previous generation Airport Extreme is the WLAN card, which I mentioned earlier is simply a PCIe x1 card that inserts into a small slot on the board.

Getting the cans off the card was easy enough, and underneath revealed two Marvell 88W8366 WLAN SoCs. These enable the previous generation Airport Extreme to support up to 3x3:3 802.11n on both 2.4GHz and 5GHz. Note that with the notation 3x3:3, the first digit is the number of transmit chains, second is total number of receive chains, and third is the total number of data streams supported.

The underside (from my perspective, really this is the top) of the whole affair is home to the device's aluminum heatsink, which consists of pillars topped with thermal pads. Those mate up with the respective chips and carry heat out.

So enough rambling on about components the old Airport Extreme, what about the new one? After taking the Airport Extreme gen 5 apart, I discovered what was changed. It turns out that the new Airport Extreme keeps the exact same revision of the mainboard (and thus the same SoC, 128MB of DDR2, 16MB flash, and GigE switch).

In fact, after I opened up my Airport Extreme Gen 5, I was greeted with the exact same board revision and marking—“820-2622-A copyright 2009.” So what is changed? The fact that Apple designed the board to accommodate a modular WLAN radio card is key.

Instead of the Marvell 88W8366 WLAN solution of the generation 4, the generation 5 switches over to a dual Broadcom BCM4331-based solution. It’s pretty obvious to see all of that too given the markings on the card “BCM94331PCIEDUAL” pretty much gives away all of the high level details—two BCM4331 WLAN stacks over PCIe x1. The markings next to the JTAG headers also give away the fact that there are independent radios for 2.4GHz and 5GHz radio. The 9 in BCM94331 usually accompanies a Broadcom reference design, though the Broadcom website doesn’t have anything posted quite yet about that official reference design, however close it is to this design Apple is using.

The BCM4331 might be familiar if you read our Early 2011 MacBook Pro review, as it’s the single chip solution being used for 3x3:3 WLAN on the notebook. Broadcom is scant on details, but the BCM4331 is a modern single chip 3x3:3 802.11n solution.

Unfortunately my attempts to remove the cans on the new WLAN card were met with failure, as they’re held on with some seriously beefy lead-free solder and the cans acted like an effective heatsink even at my iron’s highest temp.

That said, it turns out I didn’t need to remove the cans at all, as the FCC did the job. Underneath you can see confirmation of what we already knew based on just markings—two BCM4331s, power amps, and three independent RF chains per radio. Unfortunately the FCC photo isn’t high enough resolution to show what power amps are onboard, but I can make out SiGe which has been in other Apple products before.

The only other major difference (which is admittedly pretty noteworthy) is that Apple is including finger stock EMI gaskets (which appear copper colored—maybe BeCu?) alongside the outside of the metal tray which meets up with the inside of the device. There’s careful attention to this shielding around the entire tray, and much more carefully applied foil tape holding it in place near and around the antennas in the 5th generation Airport Extreme. I’m not aware of any huge interference problems with the older generation Airport Extreme, nor does it seem like including some EMI gaskets near the antennas is going to vastly improve the independence of spatial streams, but something warranted this change.

You can see the four antenna PCBs on their own positioned around the edge of the device as well. Again these are unchanged between generations.

That’s really the only change with respect to the 5th generation device. Now what about the 4th generation Time Capsule? Inside, things are much the same as they were in the previous generation. I never owned one, so I can’t compare, and most of the teardowns I’ve seen of it aren’t super comprehensive.

Construction overall is like the Airport Extreme, however, heat up the back to soften the adhesive, pry off, remove some screws, and then the top metal lid lifts off gently. There’s a fan attached to the lid which is plugged into the mainboard, however. That requires some care to unplug gently to avoid damage.

Unlike the Airport Extreme, the Time Capsule has an internal power supply, and, of course, a 3.5” HDD for Time Machine backups. With the lid off, you can already see that the mainboard has a short SATA cable, and the power supply has a SATA power connector in addition to another for the mainboard. The power supply easily lifts out, revealing the board underneath.

I suspect that some users are interested in doing things like buying a 2TB Time Capsule ($299) and bringing along their own 3TB drive instead of paying $499 for the 3TB version. I can’t speak for the 3TB Time Capsule version, but inside the 2TB Time Capsule is a relatively consumer level WD Green WD20EARS SATA II 2TB, 64MB of cache HDD.

It’s lipped with rubber and rests inside a metal caddy on three pieces of foam. On the PCB side the four mounting holes are home to some screws which mate up with a rubber support on the Time Capsule’s base. If you’re looking carefully, you’ll note one more foam piece with a cable leading off—there’s a thermistor attached to the drive using some foam and a plastic bracket.

The temperature sensor is used in conjunction with the SoC to control the Time Capsule’s fan. I played around with the drive and a heat gun (not the two together) alongside a contact-less IR thermometer, and found that the fan seems to turn on and spin at a moderate level at just above 120 F (49 C) and stays active during normal use when the sensor is around 115 F (45 C). At below 110 F (43 C), the fan turns off. At around 140 F (60 C) the fan will spin at maximum and the front LED light will flash amber, and note overheating is happening in the Airport utility. 60 C is the maximum recommended operating temperature for the WD20EARS, so this makes sense.

Google’s empirical study of disk failures has shown that between 40 and 45 C is ideal for prolonging drive life, so as long as things run in the fashion that I saw them run, it seems like drives shouldn’t fail due to excessive heat. That said, the airflow pattern in the Time Capsule seems odd—air is drawn in from the side, over and through the power supply, into the fan, then blown over half the drive, and hopefully exhausts through some small vents at the other side. Even with the fan spinning at maximum, it’s hard to really tell that any air is moving, and the only time you can even hear the fan is with it at maximum. The Time Capsule has a strong reputation for running warm, and I can definitely see why.

Upgrading the drive is simple enough. Transfer the four screws, rubber, foam, and the temperature sensor, and connect the SATA data and power cables. There’s also no need to preformat, as the airport utility has an auto formatting option as we’ll show later.

Disassembling the Time Capsule further shows some more interesting, if slightly expected details. On the reverse (top) side of the main PCB is the same Marvell SoC, switch, and flash module.

What’s different is that the Time Capsule gets double the RAM of the Airport Extreme—256MB of DDR2 instead of 128MB. Once again, we see the same dual BCM4331 based PCIe x1 WLAN card and four antenna connectors.

So there you have it, the main difference in both of Apple’s main WiFi AP products is the WLAN stack, which is a modern, more powerful BCM4331 based 3x3:3 solution. So, how has performance changed then?

WiFi Performance Testing—Methodology

To settle this, I decided on a testing methodology for comparing both generations of Airport Extreme, and the latest generation Time Capsule on two different platforms with 3x3:3 support, and a 2x2:2 design.

The first is a Lenovo X300 notebook. I acquired one of Intel’s latest and greatest WLAN cards, the Centrino Ultimate-N 6300, which is a dual-band 3x3:3 half height mini PCIe x1 card for notebooks.

It’s Intel’s highest end card, which I hastily inserted into the X300. Interestingly enough, I encountered an issue with installing just any WLAN card in the X300—it appears that Lenovo has some built in BIOS protection which prevents installation of “unauthorized” WiFi cards. Fortunately, the adjacent mini PCIe slot (ostensibly intended for cellular WAN) has no such protection. However, I then noticed another problem—the card wouldn’t turn on.

Some more searching revealed the solution—pin 20 (wireless disable) needs to be taped over to signal the card that the wireless disable switch (which doesn’t exist, since this is the WAN port) is in the on position. A quick surgical application of tape, and the card worked perfectly—take that, Lenovo security. As an aside, what a completely pointless and trivial barrier this is—the Mini PCIe standard (and moreover WiFi notebook cards themselves with U.FL connectors) are designed to be completely and absolutely interchangeable. The notion that this provides any added security (when the adjacent slot is completely unguarded) or—even more absurd, convenience—is nothing short of a surrealist notion.

My second testing platform was a 2010 MacBook Pro, which has a 2x2:2 solution provided by a BCM4322 based design. Finally, a 2011 MacBook Pro with a BCM4331 based 3x3:3 solution was my last test platform.

I tested in four locations in my home which has a loft-like two level layout and cinder block construction. My office is upstairs and in a corner, where the AP sits on a shelf about 5 feet in the air resting in its normal operating position (marked AP). Normally I mount the Airport Extreme on the wall in the vertical position using an Air Mount wall bracket, but for testing I relocated everything to a shelf for ease of changing between multiple devices and drives.

The first position (marked 1) is inside the office and on a relatively standard wooden desk. The next test location (marked 2) is in the stairway on a metal desk made out of a conductive material (likely aluminum). Downstairs are locations 3 and 4. Location 3 is on a wooden coffee table, to emulate just using the device while watching TV, and Location 4 is the most challenging position in the house in the corner of the kitchen.

The downstairs living room and kitchen area has proven to have challenging and undesirable RF propagation characteristics with the AP in my office, as the signal must propagate through two cinder block walls and change elevation. In addition, the entire staircase is one solid piece of steel, with the middle portion almost a quarter inch thick. If you draw a line from the AP to location 4 it’s pretty obvious that the testing location is directly in a shadow cast by the huge staircase.

In reality, testing is best performed in an anechoic chamber or some other completely controlled environment, but alas I lack any access to one and thus we’ll have to go with a real-world environment instead. I also did as much of my testing as possible after midnight to cut down on spurious interference from neighboring APs. Luckily I’m not in a dense urban environment and interference isn’t much of an issue.

For testing, I configured the APs to use channels 11 and 157 for 2.4 and 5GHz respectively, separate 2.4 and 5GHz AP names for manual selection, and 40 MHz channels on 5GHz. As a reminder, Apple still doesn’t allow for the use of 40 MHz channels on 2.4GHz in order to not use so much spectrum that it degrades Bluetooth HID performance.

For testing network throughput, I settled on two different things. First is Apple File Protocol (AFP) which is easy enough to test by creating an AFP share on one computer and then transacting huge files and averaging network bandwidth. Because I lack additional Macs, I turned to a positively ancient 2GHz iMac G5 with 2GB of RAM running 10.5.8 that I fixed a while ago. Rest assured, the iMac G5 does have a 1 GigE port with jumbo frame support, and ample power to serve this purpose. Instead of installing an SSD to guarantee that the sluggish HDD wasn’t throttling AFP performance, I created a 1GB RAMdisk and shared it on the network. The RAMdisk can do around 300MB/s sustained reads and writes locally and almost 800 Mbps over ethernet, so we’re not artificially bottlenecked. For a test file, I used a 500MB zip of AVCHD video and some random huge PDFs.

The second test I settled on is the cross-platform, open source iperf which is a network tool for measuring TCP network throughput among a bunch of other things. I compiled iperf 2.0.5 from source on all three Macs (yes, even a PPC binary for the 10.5.8 iMac G5), and used a version in conjunction with Cygwin on the Windows based notebook. After some tuning in my office I settled on a 2MB TCP window for maximizing throughput.

When I set up this testing, I was interested in really three things. First, the Modulation Coding Scheme (MCS) being used at each location, which tells you the data rate of the wireless link and how many streams are in use. Second, the received power level to see if I could measure a difference thanks to the increased power in the newer generation routers. Finally, actual network throughput both up and down on the wireless link.

On the Macs, it’s easy enough to get both MCS and RSSI. You can either run:

/System/Library/PrivateFrameworks/Apple80211.framework/Versions/Current/Resources/airport -I

and look at the output, or option-click on the WiFi symbol.

On Windows, getting this information is a bit more challenging as you have to rely on the WLAN card maker supplying a utility that shows any good details. Intel doesn’t directly show RSSI unless you enable logging in the advanced statistics utility and cross-correlate with samples recorded every second or so, and this proved far too time consuming. For MCS, you have to abstract backwards from the rate shown in the wireless network connection status using a table.

Performance Testing

Before I go any further, just a warning that this is dense, and even for me there are an almost overwhelming number of tables. I struggled with how to present data and settled on the format below, which consists of individiaual tables and a few graphs at the end in case you're a visual person. There's a third option however, which is the google docs spreadsheet I used while collecting data. I've made it public for your enjoyment and it contains the exact same data. (Update: Google Apps is acting wonky and not letting me share the spreadsheet quite yet—it will be posted in a little while. Update 2: Here we go, finally got the link to the spreadsheet for your persual)

First up is the received signal strength, which will show just how much difference that extra ~140 mW will buy. In reality, it’s surprisingly hard to see the difference, but again that increase in strength is only a few dBm, so I suppose it not being very dramatic here is to be expected since this is also a logarithmic scale. That said, the Gen 5 does post numbers correspondingly above the Gen 4 on average; remember that closer to 0 is higher power and thus better. The difference is actually more visible on 2.4GHz than 5GHz based on these numbers.

Next is MCS (Modulation Coding Scheme) which shows how fast the card is connecting to the 802.11n network. Here we can see how much the 5th generation airport extreme improves MCS selection in a number of cases, especially the three spatial stream scenarios on the 2011 MacBook Pro.

Locations 3 and 4, which are in challenging environments, see massive increases, previously going from the lowest possible (or not even 802.11n) rate up to much faster rates on both MacBooks.

AFS is our next test, where we transfer a 500MB zip file up and down from an AFS server and average the throughput. On the downstream side of things, the improvements aren’t substantial until we’re in a challenging RF scenario downstairs;, here the new generation wireless card in the Airport Extreme makes a huge difference in throughput on 2.4GHz, and similarly on 5GHz, though at the farthest location it’s still impossible to connect to 5GHz.

When it comes to upstream, the results are dramatic both on 2.4GHz and 5GHz. Throughput is almost always over double, thanks probably in part to the better front end and receive sensitivity of the Airport Extreme’s new wireless stack.

Iperf is finally up, which we can run on the X300 in addition to both Macs. Here on downstream the results are improved pretty substantially for the 3x3:3 2011 MacBook Pro, and across the board for the challenging downstairs RF scenarios. The same applies on 5GHz as well, and in the best case, we can push nearly 300 Mbps on the new MacBook Pro. It’s a dramatic improvement in best case throughput if you have the right card, though the Intel card ends up performing similarly on 5GHz with both the new and old Airport Extreme card. More on that in a second.

At the end of the day, the new Airport Extreme dramatically improves throughput in the best case and in a few regions where signal was previously unusable. In the worst case (location 4), performance improves from being essentially unusable to totally fine, and in the case of the 2010MBP goes from not being able to connect at all to pushing 23 Mbps.

So the unanswered question is how the 3x3:3 2011MBP manages to be much faster compared to the 3x3:3 Intel 6300 card, and I suspect the answer might be that the combination of BCM4331 on the client and BCM4331 on the AP enables Apple to use Broadcom’s frame bursting high speed modes—aka modern speedbooster. Careful observers will note in addition that while Iperf over 40 MHz 802.11n (with a link rate of 450 Mbps) delivers 311 Mbps of downstream, the same test on AFS is around 150 Mbps down, possibly due to compression. In addition, note how the Intel card lags behind in locations 1 and 2 until signal gets lower and then becomes competitive again. To me, this definitely seems to indicate some Broadcom-to-Broadcom enhancements are at play. Frankly for Apple this makes sense considering their top to bottom ecosystem control; if you have the ability to choose the card in the AP and the client, why not go with a solution that offers benefits?

Airport Extreme vs. Time Capsule

The next question is how the Time Capsule compares. Rather than re-run all 128 data-points (and then multiple tests per scenario for the purposes of averaging and removing outliers), I decided to use a subset and see whether performance and range is the same on the Time Capsule in those cases. As we showed before, the Time Capsule and Airport Extreme use the same exact wireless card, though gain is different on the Time Capsule antennas than the Airport Extreme.

For this testing, I just used the 2011MBP with its 3x3:3 radio, and ran through signal measurements and Iperf. First up is how signal strength looks.

You can pretty much immediately tell that things are very similar. On 2.4GHz and 5GHz the two are very comparable except in a few odd cases. Of course the propagation isn’t going to be identical between the two, but subjectively it’s close.

Next is MCS, and here things are again close, with the Time Capsule narrowly edging out the Airport Extreme on 2.4GHz, and things being very equal on 5GHz. It’s looking the same so far.

So what about performance now with Iperf? On the downstream side of things, the Airport Extreme comes out on top narrowly in all but the second location on 2.4 and 5GHz. But the difference is minuscule.

Upstream is a similar story, with the two being very close outside locations 3 and 4, where the Time Capsule narrowly edges the Airport Extreme out on 2.4GHz.

All said and done however, the two are incredibly close and despite the difference in gain that the FCC docs would lead you to believe, are virtually indistinguishable in some real-world testing. In my mind, if you’re concerned about WiFi performance, the Time Capsule and Airport Extreme both perform equally well.

Graphs

If you're a more visual person and find that wall of tables and text too daunting, Anand also made some awesome graphs which I would be remiss to not reproduce here for your viewing pleasure.

First up is AFS performance compared on the 2011MBP between the 4th and 5th generation Airport Extreme.

 

Disk Performance

The next question on my mind was disk performance. I don’t own a Time Capsule for backing up my singular Mac; instead I use a hard drive dock attached to the Airport Extreme over USB 2.0 for that purpose. So the question now becomes, is it dramatically faster to use the Time Capsule’s internal disk over SATA II instead of a USB 2.0 attached dock? The question is almost rhetorical (of course SATA II is going to beat USB 2.0, right?), but nevertheless we want numbers.

When a disk is attached with USB to the Time Capsule or Airport Extreme, you have the option of sharing the disk like an ordinary AFS or even SMB share with or without a password. The Airport Extreme or Time Capsule pops up under devices in finder, and then you can mount specific shares which appear just like a normal volume to OS X. Rather than time the creation of an entire fresh Time Machine backup on each configuration, I settled on measuring performance across the network to the drive using QuickBench while connected over GigE.

Before we go into the performance numbers, let’s talk about changing out drives in the Time Capsule. The nice part about the Airport Utility is that there’s a one-click erase button for automatically partitioning any drive properly for the Time Capsule. There are even options for making that wipe take place with multiple passes in case you’re trying to delete some really shady, erm, scientific research.

Anyhow, clicking erase from the airport utility makes the drive usable without the need for any preformatting. It’s almost as if Apple expects people to try to upgrade the disk on their own in the future or to make it easy to service. Doing so and then taking the drive out for inspection reveals three partitions. You can see these three partitions by running diskutil list on the drive.

/dev/disk5

#: TYPE NAME SIZE IDENTIFIER

0: GUID_partition_scheme *240.1GB disk5

1: Apple_HFS APconfig 1.1GB disk5s1

2: Apple_HFS APswap 1.1GB disk5s2

3: Apple_HFS Untitled 1 237.9GB disk5s3

So no matter how you slice it, you lose 2.2GB to overhead just from the Time Capsule partitioning things how it wants. I couldn’t find anything inside APswap, at least after a fresh format. Inside APconfig is a file named AFP.reconnect_keys which contains a mysterious unintelligible string, perhaps a hash of some kind or a token used to authenticate clients. Regardless, it’s apparent that the architecture for Time Machine on a Time Capsule differs from the external disk method when a drive is attached over USB. I learned on my own that these backups aren’t easily portable between the two, though you can actually migrate backups off of the Time Capsule to an attached disk by clicking Archive.

When using the disk attached using USB however, you have to preformat the drive as HFS+ yourself, but in that case one large partition suffices. Then you back up to .sparsebundles inside that, and essentially Time Machine treats the thing like a network attached drive.

So on to the performance testing. In true AnandTech fashion, the disk I chose to use for testing was a Vertex 3 MAX IOPS 240GB SSD, just to eliminate the disk as being a potential bottleneck in the performance benchmarking. To start, I took a baseline test with my rather generic Rosewill USB 2.0 dock attached directly to the MBP. Next, that same volume when connected across the network to the Airport Extreme using USB 2.0, then the Time Capsule using USB 2.0, and finally installed inside the Time Capsule using SATA II (3.0Gbps). I present to you the (probably) first Time Capsule with an SSD inside.

For good measure, I also tested the Time Capsule with the shipping 2TB WD Green, and one of my own personal 3TB WD Greens inside, just to see how platter drives fare. I also threw in one test where I connected over WLAN instead of gigabit ethernet directly to the devices.

First up is sequential read speeds, which aren’t critically important for backing things up (since you’re probably more concerned with write performance) but hugely important for restoring and verifying or indexing backups. Connected directly to the devices, we can essentially max out gigabit ethernet with 80+MB/s transfers, or just shy of 700 Mbps. Some individual tests are even at 100MB/s (800 Mbps), but these are the averages of those 20-100MB size tests.

SATA II here gives a huge advantage over USB 2.0 and comes close to hitting the limits of gigabit ethernet. Oddly enough, USB 2.0 performance isn’t even limited by the dock, but rather the controller. Connected directly to the notebook, performance is almost 10MB/s faster.

Reads show a similar, if interesting, story. SATA II again proves noticeably faster than the USB 2.0 route, but performance seems capped to around 40MB/s. Obviously the Vertex 3 can write way faster than this, but the controller seems to be strained to the limit when making writes. This is a bit frustrating considering mostly what the device is going to be doing is writing stuff to disk every couple of hours.

As Anand will tell you, no sequential write is ever really truly sequential, so small block performance is also important. I selected the 4 KB block test and analyzed results, which show that for writes we’re pretty much limited to the same 1.0MB/s even on a stupidly fast SSD both over SATA and USB 2.0, again the controller or network overhead with so many small transfers seems the limit.

On reads, the same story applies, with everything stuck around 8MB/s over gigabit or the dock. WLAN performance suffers considerably.

If you’re going to be doing lots of backing up and care about speed (or don’t mind the extra price), the Time Capsule’s SATA II interface makes it a considerably faster experience compared to the cheaper USB 2.0 dock solution. It ends up being twice as fast at writing and over three times as fast at reading to use the Time Capsule over the Airport Extreme. Of course if you’re going to be using WiFi to back things up, all bets are really off and the performance gains of using the Time Capsule over the Airport Extreme largely erode away.

Airport Utility and Network

The only remaining thing to go over is the Airport Utility and general network functionality. I’m not going to go into very much detail at all about the network side of things (routing performance, etc.) since honestly nothing has changed here—the board and SoC are the exact same.

To start, all configuration is done through Airport Utility on OS X and Windows. There’s no web-based portal like just about every other attached network device. On all of Apple’s WiFi products, Airport Utility is the exclusive point of configuration. It pops up a list of devices, including unconfigured ones without the need to connect over ethernet. Admittedly that’s a nice touch that makes configuration super easy.

There’s either the choice to have a wizard set everything up, or you can do it manually. The first page is summary, which lists some basic information, status, and high level information. Status mirrors the front panel LED which either glows green or amber depending on whether things are working fine.

For example, when the Time Capsule starts overheating the LED blinks amber and pops up an alert. The utility can also monitor for when problems happen.

The rest of configuration is honestly pretty straightforward. Each of the radios (2.4 and 5GHz) can be set to its own SSID, there are a wide selection of radio modes and the ability to set channels manually. As an aside, if you’re going to set the 5GHz channel manually, select one over 149 to get maximum power and avoid dynamic frequency selection chanels in the US.

There’s honestly everything you’d want from a consumer level router or wireless AP, and a couple of extras like RADIUS server support and guest network creation.

If you run the devices as a router, you can also port forward, configure the DHCP server (including static addresses assigned to specific MACs), and all the usual fixings. That said, some of the options are a bit barebones compared to more enthusiast open source packages, but it gets the job done.

In addition to sharing disks, the Airports can also share printers attached using USB. I’ve experimented around and even a USB hub plugged in with devices attached to it works, so you’re not limited to just a single port.

One of the features Airport Utility has that a lot of other consumer devices don’t is a nice way of viewing the status of attached wireless clients, including a little graph that continually tracks. The utility will show each client MAC address, signal, noise level, rate, and what network mode they’re using. It’s an incredibly useful diagnostic tool that’s honestly a step beyond what Tomato or even DD-WRT report for clients.

The Windows version of Airport Utility looks a bit like a program from the Windows XP days, which is about par but not unforgivable. Everything inside works the same way, except this version forgoes the client signal graph and instead just presents a table with data. You can still do everything else, however.

Update: Our own Saumitra Bhagwat pointed out to me that on iOS 5 Beta, you can actually use an Airport Utility (lite) built into the WiFi settings page to configure a new Airport Extreme or Time Capsule. You have to be in the initial setup state (reset completely to defaults) to see it, but when in range, a "setup an airport base station" window appears, allowing this lite level of configuration. Depending on your existing setup, it'll offer some suggestions as well (for example, as I'm attached to a current Extreme, it offers to extend by default).

Another small but useful thing is that both devices support SNMP, so you can do stuff like track traffic on all the internal interfaces on the device using software like MRTG or Cacti (both of which I’m a huge fan of). There’s even a MIB for how many wireless clients are joined. If you’re not into using SNMP yourself, there’s a simple utility out there called Airport Flow that gives you a per-interface graph and the total number of clients attached.

 

You need to specify the device IP address and what interface to monitor, so you might have to snmpwalk the airport (snmpwalk -c public [ip addy]) and look for the line with each interface name:

IF-MIB::ifDescr.1 = STRING: mgi0

IF-MIB::ifDescr.2 = STRING: mgi1

IF-MIB::ifDescr.3 = STRING: bwl0

IF-MIB::ifDescr.4 = STRING: bwl1

IF-MIB::ifDescr.5 = STRING: lo0

IF-MIB::ifDescr.6 = STRING: wlan0

IF-MIB::ifDescr.7 = STRING: wlan1

IF-MIB::ifDescr.8 = STRING: bridge0

But there’s nothing complicated about that.

Honestly, for routing and firewall, the Time Capsule and Airport Extreme get the job done but without much flair or ability to tweak low level things like I’d like to see. I generally use either a pfSense embedded x86 box, WRT54G-TM running Tomato, or a WRT-600 running DD-WRT for NAT routing, firewall, and DHCP myself, just because those options offer a ton of fine grained control over the network side, and then an Airport Extreme for WLAN. That said, the Time Capsule or Airport Extreme can do all of that in one box.

Conclusion and Final Thoughts

We’ve gone into extreme detail about the Time Capsule and Airport Extreme updates. I originally intended to have this posted with a hint of irony on WiFi day (8.02.11) but instead ended up spending that special day doing more testing and running even more instances of Iperf to make sure our numbers were solid. What’s changed between both previous generations is simple—the Time Capsule gets an official 3TB option, and both the Time Capsule and Airport Extreme now have a much more powerful, modern, and better-performing BCM4331 based WiFi stack.

The result of the move from Marvell to Broadcom is twofold. First, performance and range is definitely better thanks to more transmit power and the improved sensitivity afforded by newer generation chipsets. Second, the combination of lots of Broadcom in Apple’s hardware lineup (from the iPhone, iPad, and MacBooks, to iMac and Mac Mini) with Broadcom in the access point likely allows for the use of frame bursting or some other packet aggregation technique that speeds things up in some scenarios. It’s another example of how having that complete hardware control can in fact result in some benefit—in this case, faster WiFi.

Before this update, there were so many rumors about both iOS based Airport products, that the Time Capsule would cache software updates locally, and that the whole thing would somehow tie into Apple's iCloud solution. None of that exists right now, and it's looking like (for now) the rumor mill has some egg on its face. I waited patiently for Lion to launch and half expected things to turn on and render half of the review null; instead that day came and went without much change at all. As of this writing, the core functionality of the Airport line remains the same as it was before—sans any iCloud/iOS magic/local update caching.

There’s a stigma that Apple gear is more expensive, and for the 3TB Time Capsule that may be the case, but the Airport Extreme is actually right near where it should be. Take for comparison the Linksys E4200, which is a 2x3:2 device on 2.4GHz, and 3x3:3 on 5GHz, and also Broadcom based. That device runs for $179.99 and features similar functionality including a USB 2.0 port for sharing devices. At $179.00, the Airport Extreme offers full 3x3:3 on both 2.4 and 5GHz, albeit the E4200 does have considerably more Tx power, which we'll investigate in a forthcoming article.

I guess the reason that I personally use an Airport Extreme (in conjunction with another device for NAT) is that it's really one of a small number of 802.11n dual-band APs I've tried that actually works without locking up, becoming unstable periodically, dropping the session from overheating when being pushed to 100% for hours, or requiring a daily reboot. There are just so many other consumer level 802.11n APs that either fall short or are incredibly frustrating and unreliable. Thus far, I've been using an Airport Extreme Gen 5 and Time Capsule Gen 4 as my primary AP with over 12 devices attached to each one for the greater part of a month without a single instability. It's that kind of stability that really sells it for me, even with 3x3:3 out of the picture.

That kind of sums up WiFi in general—ideally, it should work and be something transparent to the user instead of a constant consideration. I wager the vast majority of Airport Extreme and Time Capsule owners have no idea what 3x3:3 is or how to even check their physical link rate, and for the most part that's a testament to how stable these devices are. Maybe that's the reason why Apple doesn't make a super huge note about changes like markedly improving their WLAN connectivity. One thing is for certain, Apple's wireless division is either playing it incredibly cool, or honestly not getting the credit it deserves.