Nick Ali, author of the Ubuntu Weekly Newsletter (among other things), writes that Ubuntu doesn't get a fair shake in a Pingdom report on update site availability. Microsoft had a measured 100% response to pings, Apple a 99.9%. On the subject of Ubuntu it says:
Ubuntu on the other hand came off worse, with only a 98.64% availability for its main repository. That is a total of 1 day, 5 hours and 45 minutes in the three months of this survey.
This brought out an ugly side of Ubuntu Marketing: the self-appointed Defenders of Ubuntu, ready to strike down all perceived slights against it in forums, blogs, and wherever else informed discussion rears its ugly head. Nick Ali (not a Defender, and out of character today) is apparently upset that the headline is misleading (would "Microsoft, Apple trounce Ubuntu" be sufficient?). 'Ed Vim' comments to the Pingdom post itself, saying:
I find this article to be a bit misleading, not only in the title but also the content. The title is just another headline grabber for other sites to use in the typical MS FUD campaign. And let’s face it, a plain vanilla Ubuntu or OS X system with no updates will still be safer and more functional to use than a clean Windows install. This ‘alleged’ update situation is a very minimal issue, As for content, just read the previous comments.
Ubuntu lives in a glass house. You can see nearly everything we do (even the nasty comments on general computing sites). So ask yourself this-- who reports network availability to Ubuntu and why hasn't anyone cited them yet? I have no doubt that Microsoft and Apple have someone looking at this and making sure it's up to par.
People are making out like its unfair that Ubuntu is penalized for a downtime on its archive because of a new release. I agree in one sense, but not completely; the time period over emphasizes the release week and would have been more relevant to Ubuntu users, current and potential, if a six month or year long period were measured instead of the one month coinciding with a new release. But we're only talking about doubling the length of the study. Assuming that the entire outage was contained to release, we'd still be lucky to get two nines.
Still, I don't mind the comparison. Rather, we should feel honored to be chosen as a representative of Linux as a whole, and try to live up to such expectations. One thing not spoken thus far, is that a separate mirror exists for security updates, hopefully so that exactly this problem doesn't spill over into slower security. It would be interesting to see how that archive matches up with the general archive.ubuntu.com availability, but currently both seem to point to the same data center.
And we do have serious downtime problems during this time period that no amount of amortizing can undo this semi-annual anomaly. Mirror admins jokingly compete to see who's got the biggest pipe during release weeks. The main archive is known to be unusable during this time, and the website itself typically reverts to a low overhead, high information view to compensate for the rush. Some clever people decide to upgrade a few days before release to avoid the rush (if only they'd do it before freeze, we'd have more testers!). If we seriously intend to fix Bug Number One we need to hold ourselves to the same level of accountability that we do Microsoft or Apple. The existence of unofficial mirrors isn't an excuse, and in fact may be a security hole. We have talented people and source code, but without the willingness to accept how we look in the mirror as truth, we cannot truly improve.
So the next time you encounter criticism of your favorite software, I ask you to please take a moment to ask yourself two questions: Is it true? and How can I make it better?