As i-boy notes, Twitter failed big time this morning leading to quite a bit of disappointment from those hoping for real-time updates to their handset around the globe, especially those who weren’t able to ‘tune in’ to the likes of the MacRumours live offerings.
The Twitter guys posted a brief note:
Link: Twitter Blog: MacWorld
Twitter is currently experiencing some slowness related to the massive number of updates around Steve Job’s keynote at MacWorld. We’re working on it and will keep you updated!
Sending 140/160 character messages is not difficult. It’s a piece of simplicity. I speak from experience. We were processing hundreds of thousands of messages in short periods with our Impulse text-to-screen nightclub service regularly. It’s all about smart and effective code, efficiently tuned databases, powerful servers and good tier-1 prioritised connectivity. Then all you can do is wait and hope that you’re not a victim of overloaded cellular transmitters or network traffic — problems that are really out of your hands.
It’s a little unfortunate that Twitter wasn’t able to cope with what one imagines was a rather small traffic peak (given the huge volumes Twitter generally handles).
Biz from Twitter posted later:
Macworld is only one event, in one city. Twitter must be reliable around the world, around the clock, and it must accommodate all sudden bursts of real-time messagesâ€â€everything from Apple announcements to natural disasters.
He wanted to point out that they’ve hired a chap by the name of Lee to oversee engineering and operations. His first goal?
to help turn Twitter into a reliable communication network that people all over the world can depend on every day
Totally. I’m sure they’ll sort it out. Reliability with this kind of service is key, especially with the early adopter geek audience that must comprise a high percentage of the Twitter user base. If you can’t stay reliable, it’s just 15 seconds and one group email to swap your network to any number of competing offerings.
The problem with things like this is uneven capacity. The reason the mobile networks coped so well during emergencies like the tube bombings is because they’re massively over-engineered to cope with 3 yearly spikes in traffic. Xmas Day, New Year, Valentine’s Day. The rest of the year that capacity is sitting idle – and that’s expensive.
I don’t use Twitter (I know, I’m a Luddite), but does its business model justify a massive amount of hardware which is only going to be used once a year at MacWorld?
You could apply that viewpoint to the mobile operators too, Terence. I think they need to invest effectively — particularly if you setup a pay-on-use model with your suppliers. That said it’s all, to an extent, irrelevant if there’s insufficient money to fund it all in the first place.