Just how well did Engadget do last Tuesday?
I don’t like to talk about Engadget’s traffic, I think it’s kind of gauche. I remember when I first started at Engadget in June of 2004, I worked for free and we were doing thousands of pages a day. And I was absolutely ecstatic with just that. It was the peoples’ blog about tech news — we wrote, and our readers connected via comments, emails, etc. Obviously some things changed; we got a lot bigger in terms of readership, and in turn editorially to keep up with the demand for Engadget content. But Peter and I never wanted it to seem like we we disconnected and got too big for ourselves, or that we weren’t your blog anymore. It’s easy to be intimidated by big numbers like 10 million pages a day, so we don’t talk about traffic, I just don’t allow it.
Business 2.0 published a short piece on Tuesday’s traffic, which led to a Calacanis quote and a Tech Crunch piece and, of course, a digg hit. So I guess now’s a good time to clear up any misconceptions there may be. First: we DID NOT go down — not for 7 seconds, not even for a second. (We sure as hell did go down last time at WWDC, but not this time. More on that in a sec.) Peter’s quote to Business 2.0 was (and I was there), “We didn’t go down, but if we did, it was at worst 99.9% uptime.” So to be clear: I checked again later with our techs, and we didn’t go down at all.
Which brings us to misconception number two: this 10m page number. It’s not 10m pages, it’s actually more than 10m but hey, if we’re going to “be numeric” as Gates says, then let’s. Again, we quoted that traffic was “into 8 figures”. It was pretty well into 8 figures, and counting the rest of the Engadget network (Mobile, HD, Japanese, Chinese, and Spanish) and that’s even more still. I’m not going to discuss numbers, but I was very proud of what we accomplished, and I’d have been proud if we only did 10m.
The next misconception: we didn’t have increased uniques, just increased traffic. Not so. Looking at our uniques count, we were between 300-400% that day. Our page view traffic between the hours of 12 - 2PM EST (the hours Jobs was on stage) accounted for roughly only 23% of the day’s traffic. In other words, people sought out the news at Engadget. (On a sub-misconception, almost none of that traffic was because of AOL. That day less than 1% of our traffic was referred to from AOL properties — glean from that what you will.)
And finally, the last misconception: we only stayed up because we are on a huge server farm in AOL-land. Totally untrue. The crew at WIN have built the best and most scalable blog platform I’ve ever seen. That’s not kool-aid I’m sipping — it has to be this way because they designed it to stand up to what happens when Peter and I blog Jobsnotes. Since late 2005, any time Engadget has gone down under load — which has been only a handful of times, one of which being during a Jobsnote — it’s been because of something non-Blogsmith related (i.e. a switch died, someone didn’t prep, etc.). From what I was told, our server farm, which usually operates at under 1% total load during normal traffic, didn’t exceed 20-25% load. Which is why we have stayed live under similar load with half as many servers in the past. Honestly, really, it’s the platform!
So did Gizmodo go out? I honestly have no idea, I was blogging the entire time. But I heard from people that they were inaccessible at times. Not that that really matters because it isn’t a pissing contest. People should realize that a Jobsnote is like the worst DDoS any site could ever imagine. As a former network tech, if Engadget went down on Tuesday I would be upset but I’d certainly understand why. The Gawker people have obviously done a great job on their stuff too, and up or down, they did better at staying alive during than just about anyone — just not us (and, from what I hear, Macrumors).
That was definitely an interesting read.. and I wasn’t even aware Engadget until that day as well.
Comment by Eric — Sunday, January 14, 2007 @ 12:41 pm
Out of curiosity how many servers do you guys have? If you only hit 20-25% load on the iPhone day I assume you have quite a few.
Comment by Ryan Wagner — Sunday, January 14, 2007 @ 12:56 pm
It’s always a good thing to read things direct from the horses mouth — err, blog.
Comment by Ronald Lewis — Sunday, January 14, 2007 @ 1:16 pm
We don’t have any dedicated servers. Our cluster serves all of the Weblogs, Inc. network, TMZ, and the other AOL properties which use Blogsmith (our platform). I don’t know exactly, but I don’t think it’s more than 50.
Comment by Ryan Block — Sunday, January 14, 2007 @ 1:19 pm
I am not a Gawker employee, but as a reader who was refreshing Gizmodo during the keynote, they appeared to stay up for most of the time. There was a blip about half-way through, but nothing compared to the “oh fuck, we’re down for a couple of hours” of the past. (It was less than a couple of minutes by my single-browser-refreshing estimation.)
The upshot of all of this is that it seems like all the big blog networks are finally getting their shit together, performance-wise. Good on Engadget for holding up under that blast of traffic. That’s no minor accomplishment.
Comment by Joel Johnson — Sunday, January 14, 2007 @ 1:34 pm
[...] read more | digg story [...]
Pingback by Just how well did Engadget do last Tuesday? « Universe_JDJ’s News Blog — Sunday, January 14, 2007 @ 1:51 pm
Hey Ryan, good post. Very honest, per usual. To answer your question, we didn’t go down, but the multi post format sucked in this case for the live blog. For lack of sleep, we forgot about usability. Duh — we would update a single post, with timestamps, and then jump to another when we felt the subject deserved a headline. Impossible to follow, really. A big mistake, but that’s the difference between my 6 months of experience in this game, and you and Pete, as vets. Live and Learn.
Fantastic week for both sites, though, that’s for certain.
Cheers,
B
P.S. Holy F*ck, OVER 10 million? Maaaan.
Comment by Brian Lam — Sunday, January 14, 2007 @ 2:00 pm
[...] UPDATE: Ryan Block from Engadget says that their site did not go down. Congrats to them for withstanding that much traffic! [...]
Pingback by futurekarma » 100% Uptime Indeed — Sunday, January 14, 2007 @ 2:13 pm
10 million pageviews, it’s really amazing.
Congratulations, Ryan!
Comment by Nick — Sunday, January 14, 2007 @ 2:59 pm
Yeah Brian, good game, great week! You guys still stood up to the beating, not many sites can say that. I’ll see you soon!
Comment by Ryan Block — Sunday, January 14, 2007 @ 6:38 pm
I think Gizmodo did far worse. There was big confusion initially since everyone expected them to update the same page with minute by minute reporting but what they did was break up the keynote into several individual posts. So I (and I think several other readers like me) moved to Engadget when someone in the comments mentioned that you guys have a much better coverage.
Comment by Manu Sharma — Monday, January 15, 2007 @ 12:44 am