More Commentary on Our Wireless Woes

I’m still in the midst of my dissertation writing—the first draft is in, but there’s a lot more work to be done—but, since there were some questions about our wireless problems, I thought I’d take the time off to respond. Nothing new learned, but a few data points, gripes, and commentary.

  • Something funny is happening inside Apple with this bug. They’ve known about it since this summer, as is evidenced by the number of posts in their Knowledge Base. We opened a ticket and, with Aruba’s help, gave them a bunch of data in October. I’ve also opened at least two tickets through Apple Education and one through In all cases, everyone in Apple is really helpful and escalates the problem to an engineer who seems 1) surprised and 2) genuinely interested. Then all these cases fall off of the map.
  • The lack of follow-up makes me think that either it’s being quashed inside of Apple or else their internal lines of communication suck.
  • Aruba had to be reminded of the unresolved nature of this problem, but is in our corner again pressing Apple. We also have a new Apple Engineer in the NYC area that’s taking up the issue. Of course, it was news to him when we brought it up last week.
  • Posters on Apple’s knowledge base who have the know-how and resources to test this carefully find the same thing that we do: the problem surfaces in 10.4.10 and isn’t fixed in 10.4.11. It also doesn’t manifest itself in 10.4.9, Linux, or Windows XP on the same exact machines that give problems with 10.4.10/.11. This means that it’s a software problem, not a hardware defect.
  • I’m beginning to think that the problem is with an OEM driver supplied
    by Atheros, over which Apple has little direct control. This is pure
    conjecture, but might explain both the appearance of the problem in 10.4.10, and Apple’s amnesia on the matter.
  • The machines that exhibited better behavior with the Airport Extreme Update 2007-004 patch mentioned below have since gone back to misbehaving.
  • We’ve had some marginal luck replacing the Aruba 61 access point in our office with an Apple Extreme with a different SSID. I think that this is a fluke, since I’m convinced that the problem is on the computers, not the network.
  • The last two points, plus the fact that we have good days and bad days, good areas and bad areas, makes me think that environmental factors exacerbate the underlying bugs. This would explain why others don’t necessarily have the same problems that we do.
  • I think that the environmental factor that exacerbates the issue is the high number of interfering wireless networks in our area. We’re in Manhattan, and have dozens of neighbors with wireless networks visible from any location.
  • Upgrading to 10.5.1 seems to solve the problem. However, we haven’t yet installed it on enough machines to know that this perception is not also a fluke.
  • PPC machines seem much more stable than Intel machines. The worst affected machines are 2007 MacBooks, but our 2006 MacBooks are also flaky.
  • Some machines, even of the same batch, are flakier than others. This may be due to different usage by their owners, or differential interference from other networks (where the users usually work on their computers), or different expectations and reporting by their owners.

Overall, this issue is really giving us a black eye. Our Director of Technology thinks that it’s an Aruba problem, and I’m catching heat for that, since I pushed for them over Cisco or a third company (who’s name escapes me at the moment)—so, naturally, I’d like this problem resolved. Also, since we’re planning on expanding our 1-to-1 program next year, we really need a rock-solid wireless network. If 10.5.1 is the answer, fine—but it’ll take time to win back our users’ trust. And I really could have used the time that I’ve lost on this mess to take care of other projects.

Update 2/17/08
Well, we’ve finally gotten enough users on 10.5.2 to glean some reliable information. Unfortunately, it looks grim: For us, 10.5.2 ain’t the answer. It may be marginally better, but our users still drop connections to our email server (FirstClass). Since that’s the single most important app—and particularly intolerant of hiccups on the network—we’re still left without a good wireless solution. Rats.

2 Comments Add Yours ↓

  1. Aaron #

    Thank you very much for your detail. it is extremely helpful as we enter the process of buying a new wireless system. I hope a solution is found quickly for your sake. I know how hard it is to combat the perceived cause of the problem.

  2. Tom McNicholas #

    10.5.2 is out… fixes AD binding of clients and other goodies..

Your Comment