20 Comments

Summary:

Updated: Users of social networks choose where to spend their time based on factors entirely outside of those such as uptime and reliability, according to report issued Tuesday (PDF link) by Pingdom, a service that tracks web site uptime and optimization for companies. Not that such […]

pingdom-small Updated: Users of social networks choose where to spend their time based on factors entirely outside of those such as uptime and reliability, according to report issued Tuesday (PDF link) by Pingdom, a service that tracks web site uptime and optimization for companies. Not that such things aren’t important — after all, a social network isn’t going to be of much use if people can’t log in or use the features. But the Pingdom report shows that when it comes right down to it, those things don’t matter nearly as much as one might think.

Take a look at the chart below, which sorts social networks according to their total downtime in 2008.

downtime

Notice who’s up there near the top? A bunch of networks you likely never use (or at least not very much), including Xanga.com and Classmates.com, as well as Imeem and MySpace (although Pingdom admits that its data for Imeem was incomplete). And right down there at the bottom in terms of reliability is…yes, you guessed it: Twitter. The social network that is currently growing like a weed on steroids — the one that everyone is talking about — had the worst uptime record by a landslide: Its downtime in 2008 stands at more than 84 per cent hours, or almost twice its nearest competitor, LinkedIn.

Here’s another eye-opener. This chart shows Twitter’s downtime per quarter:

twitter-quarter

Update: That’s right — nearly 50 percent of Twitter’s downtime took place in the second quarter of last year. Twitter was down almost 50 per cent of the time. That is a pathetic and dismal record by almost any measure. There are apps and services that are still in the private alpha stage with uptime records that would put Twitter to shame. But what is the social network that everyone wants to use? I’ll give you a hint: it’s not Xanga (no offense to you Xanga-ites). And one of the big reasons for Twitter’s downtime, of course, was the simple fact that it was trying to scale quickly enough to keep up with its growth (that, and it arguably had the wrong kind of infrastructure to start with — but that’s a different story).

When I wrote about Twitter and its downtime last year, I got a fascinating comment from someone who ran a small technology startup — not a Twitter knockoff, but similar in many ways. They spent a lot of time and money building a great infrastructure, robust and scalable, with all kinds of features. And what did it get them? A great service, with very few users. This person said that he would much rather have had Twitter’s problems: lots of downtime and scalability issues, but also lots of devoted users. Uptime isn’t much good if there isn’t anyone around who cares whether you are up or not.

  1. You’ve completely misinterpreted the (rather confusing) graphs. What the report actually says is that Twitter’s overall uptime for 2008 is 99.04% (see p8). The “Downtime per quarter” graphic you have shown just indicates that 48% of all Twitter downtime last year was in Q2. In fact you can see from the previous graph on p8 that the total downtime in Q2 (April-June) was around 40 hours. Based on 91 days per quarter, that works out at a somewhat less frightening 1.8% downtime (98.2% uptime).

    Note: the only one of these I use is Linked In.

    Regards Nigel

    Share
  2. Thanks, Nigel. I’m glad you noted that they were confusing :-) I will clarify.

    Share
  3. I don’t think you can make the conclusion that uptime doesn’t matter based on these number. It should come as no surprise that services with less users have better uptime and vice versa.

    Share
  4. Thanks to everyone who pointed out that I read the graphs incorrectly — the first one is hours of total downtime, not percentage, and the second one shows percentage of total downtime in each quarter. That’s what I get for reading a report like Pingdom’s late at night, and too quickly. I think my point still stands, however — Twitter has the worst record for downtime by a long shot, but is still one of the most popular social networks.

    Share
  5. [...] the rest of this post at GigaOm) [...]

    Share
  6. [...] in the New York Times, but it comes from Matthew Ingram at GigaOM.com.  The article can be found here.  The Pingdom study it discusses is [...]

    Share
  7. I completely disagree with you, Mathew, on two major points.

    First of all, for the most part, the services listed above with the lowest amount of downtime are generally the most popular. Facebook, Myspace, Imeem (you may not use it but it’s more popular than twitter) are all near the top. The general data seems to contradict your assertion.

    Secondly, twitter is not as big as you seem to think it is. Twitter is a very popular site, but its traffic is one of the lowest of al thel networks you have listed. This would once again seem to contradict your assertion and indicate that uptime *is* important.

    Share
  8. Thanks for the comment, SD. I didn’t say Twitter was the biggest — I said that (in my opinion) that it’s the one people are talking about the most and the one that is growing the fastest. I certainly haven’t read or heard that much about Xanga or Imeem in that context, although I don’t know offhand what their growth rates are like. And Twitter is growing quickly despite having the worst uptime — that’s my point.

    Share
  9. I think this is a great insight into the social network arena. If a service proves innovative and revolutionary, user adoption should arguably outweigh company resources. In other words, services and social networks like Twitter should strive to achieve their “breaking point.” Twitter’s ‘fail whale’ is a testament to their unbelievably growth.

    http://www.techsoomer.com/twitter-teaches-startups-hunt-fail-whale/

    Share
  10. [...] Information source : gigaom.com [...]

    Share

Comments have been disabled for this post