4th January, 2010
My job means I frequently see Twitter news and details within seconds of it breaking; and more often than I would like, I am left with my head in my hands.
This is because I have seen way too many blog posts from people who really should know better, making mistakes when it comes to what Twitter’s up to.
Let me explain.
Most websites will balance the load that they are experiencing across the resources they have available. This keeps all the servers equally loaded and ensures users get a consistent experience; it also stops one server getting overloaded and dying, and starting a chain reaction.
However, there is only so much you can do with how many servers you have.
So when Twitter is experiencing unexpectedly high levels of load - more than their servers can all cope with - they turn off more advanced features in favour of keeping core services running.
So when you see that the Retweet functionality may not appear, or that Lists have temporarily disappeared, this is almost certainly temporary, and just to ensure that these high-complexity services don’t interfere with the core running of the service - the actual tweeting and the APIs.
The modularity of this approach is excellent, and means that despite its massive volume of data transfer and millions of people expecting it to operate in real-time, Twitter can still handle this under epic load.
Here are some examples from some of the most well-known “social media” journos of how not to do Twitter coverage when something like this happens.
Bonus points to Mashable’s authors; I couldn’t find a single post where they’d made this mistake.
Let’s learn from this.