An analysis of the “slashdot effect” – page 2
I woke up on the morning of July 13 to the sound of a new email notification. This isn’t unusual, it happens almost every morning. I got up and checked my email. Over 100 emails — which is very unusual: new comment notifications, emails from nagios telling me that Apache had kicked the bucket a grand total of 74.6 times, pingbacks from other blogs. I wondered what the hell was going on, it seemed the world was ending somewhere up in Canada.
The worst possible thing had happened: I had mirrored the BBC Beethoven symphonies, posted them to my k5 diary, my blog, a discussion forum, and a very buried comment in another slashdot discussion, and somehow they’ve found their way to the front page of slashdot. Now, they say that slashdot will post anything, and this actually isn’t the case. I’ve tried to get things I’ve done/written posted to slashdot, and it hasn’t worked out. But go figure, the one time I didn’t want something there, it ends up being posted. I was in the middle of a full-bore, front-page slashdot beating and there were untold numbers of readers attempting to download ~350MB of mp3s from me simultaneously.
Under normal circumstances, I would be happy about the mention, because all writers like readers. But not when I’m mirroring half a CD’s worth of music.
Well I poked around, and I couldn’t figure out what nagios was bitching about. My site was still responsive, if a little sluggish. I then discovered that my mirroring of the mp3s was illegal, which I didn’t know. My first thought was “thank God I have a reason to take them down,” and my second thought was “Damn, now when everyone comes looking for the symphonies, they’re going to be disappointed.” I was conflicted, but I was a good little Internet citizen and took them down, mostly because I don’t like copyright infringement, but at the same time, it is cool to get four metric buttloads of traffic.
Anyway, I took the files down, but not before some impressive numbers were put up, thanks to slashdot. Apparently it was still July 12 according to my server when the story was posted, but most of the traffic is from the 13th.
Bandwidth usage/Page impressions:
- July 12: 12.73GB/1,755
- July 13: 167.39GB/16,082
- July 14: 756.88MB/1,517
The 167GB isn’t particularly interesting or surprising until you consider that the bulk of that was done between the hours of 1am and 7am EST — before most of the East coast is even out of bed. So that’s largely the UK and the Continent. Had I left them up, I probably would have done somewhere around 5TB of traffic (16,000 x 350MB = ~5TB). That’s a little much. 16,000 people downloading is a very real possibility as smaller bloggers pick you up and their readership finds you. These residual effects were particularly noticeable as more and more bloggers found my mirrors for the high-def Episode III trailer — which was posted on slashdot to begin with where it propagated down to smaller sites over time. From a purely sociological standpoint, analyzing the logs for the Star Wars trailer would provide a more interesting route of propagation over time because it had the first initial slashdot rush (similar to the Lego video), and then it propagated to many smaller discussion forums and blogs, resulting in more sustained traffic after the rush than the BBC symphonies and Lego Star Destroyer video. Had I kept the symphonies up, there is no doubt in my mind that a similar route of propagation would have taken place, on a much larger traffic scale. I will probably write about this in the near future.
The MRTG graphs from July 13 are obviously more impressive than they were when I mirrored the Lego Star Destroyer video:
Ultimately, it’s a case of “Slashdot giveth and slashdot taketh away” — the blessing and the curse of the small Internet publisher: the day after I was linked on the front page, my PageRank increased from 3 to 4. For many smalltime publishers and bloggers, it’s a case of “at what cost?” Is using 300GB of bandwidth a smart way of increasing PageRank (and ultimately) readership? If you’ve got a hosting plan that supports it, sure. But what about those times when perhaps you don’t want 17,000 people downloading 350MB of music from you. Granted, most sites that are slashdotted are not mirroring large files, but they are on shared webhosting plans where they’re sharing a server with 200 or more people, and which cannot cope with almost 20,000 visitors in a 24-hour period. All the same, many of them are happy when slashdot arrives because it can be an exciting time in an otherwise dull world of smalltime Internet publishing.
A random note in closing: the total number of page impressions for just these first two weeks of July is approaching the total number of page impressions for rianjs.net since its inception. And it’s all thanks to Slashdot.
Conclusion: Slashdot should capitalize on their unique position by selling stickers that read “I survived the Slashdot effect” — every dork in the Universe would buy one. Including me.
< < Previous page | polyscience.org home ^^
No Comments »
No comments yet.
RSS feed for comments on this post.
| TrackBack URI
You can also bookmark
this on del.icio.us or check the cosmos