March 9th, 2008 Richard
I have been a practitioner of mindfullness meditation for some years now, and, one of my favorite teachers is Echart Tolle. I have heard/seen a number of audio and video recordings of different talks he’s given. So I was intrigued when I heard about a 10 week live web class that Oprah Winfrey was doing with Echart Tolle, from my wife, who teaches meditation. The class uses his most recent book, A New Earth, as the text. This whole thing is really cool to me because: a) It is something that would not be possible without the World Wide Web; and b) The web is being used, in a big way, to promote a positive change in the world.
So we sat down to watch the first week of the class, and it seemed to go fine at first, but after a short period of time, it started slowing down and stalling and, finally, pretty much stopped all together. All of our friends who tried to watch, and apparently many people around the world, had the same experience. Of course, most people attributed this to the half a million people sucking up bandwidth interacting with the show’s servers – maybe the web just can’t handle this sort of thing.
… there were up to about 800,000 users when a logical error in the caching servers caused the system to crash. It is important to understand that the only way to ever find a coding error like this is to put a system truly under stress. You can’t simulate 800,000 users in the lab, you need to play with live ammo. Kudos to Oprah and her team for pushing the envelope this hard.
The crash was not caused by a lack of bandwidth, an overwhelming number of users or any infrastructure issues at all. It was a simple coding error – nothing more. The error was identified and is now fixed. It is entirely possible that next Monday’s webcast will enjoy over a million users, maybe more.
Whew, that’s a relief. Well, I’m not too sure about he accuracy of the report, but I will say, the same thing I said last week when the class stopped, which was , whatever happened, Oprah will make sure it’s fixed by the next class (tomorrow), and we’ll be tuned it to see what the web can do.
Update: I’ve learned that, according to Information Week, the initial problem was that the web servers were overwhelmed with “throughput demands of 242 GBps”.