So much for scaling. Steve Jobs’ keynote at MacWorld today brought Twitter to its virtual knees. The little guy’s web app is only just beginning to recover, while Twitter clients seem to be confused still about the number of requests I’ve made, meaning the Twitter API is borked too.
For those who care about Twitter, is this an issue for you? Will you leave Twitter for Jaiku or Pownce? Will you hope for the DataPortability.org knight in shining armor to rescue your network from janky old Twitter?
Puneet has some thoughts, and incidentally, let’s welcome him to the blogosphere. You may recall Puneet’s piece on life in the bullpen; I’ve been encouraging him to bring his viewpoints into the Wild, and he finally has.
I don’t care about Twitter and its downtime, and you shouldn’t either. Twitter is extremely useful as a network to me, but it’s not mission-critical. If you think it is, wait longer.
As an aside, Twitter is built in Rails and uses Amazon’s S3 for storage, so do Twitter’s failings reflect poorly on either of them? Or do you consider the downtime woes to be the result of poor development and/or planning on Twitter’s part?
Is it coincidental that Jaiku announced scheduled downtime for “maintenance to improve performance” the week after new parent Google joined DataPortability? Twitter has proved the concept of “micro-blogging”, and with Google’s massive infrastructure, downtime would shrink to hours over a year, not days. Throw in the ability to migrate some or all of your Tweeters, and the last one out of Twitter, kindly outen the lights.
My guess is Tweeters have very low loyalty to Twitter, otherwise why so much grumbling about the downtime of a free (as in beer) service without any ads? Plus, Twitter has no business model or way to keep its users. The only lasting impact Twitter may have is on the geeky lexicon; tweeting may stick as a verb, but I doubt a Googlized Jaiku will have much trouble replacing it as the de facto micro-blogger.
Today may be Twitter’s tipping point, the day the tweeting died.
Interesting thoughts. I’ll leave it to the more technologically gifted to make the predictions and to explain the tech, but like you said, who can complain about a free glass of beer without ads?
Should Twitter go the way of ICQ, CompuServe (my first online community), or others, it will be interesting to see. In the meantime, my PDX “friends'” tweets have opened my eyes to a whole new world of OpenID, dataportability, and the real value of social/business networking.
http://twitter.com/gwalter
Interesting thoughts. I’ll leave it to the more technologically gifted to make the predictions and to explain the tech, but like you said, who can complain about a free glass of beer without ads?
Should Twitter go the way of ICQ, CompuServe (my first online community), or others, it will be interesting to see. In the meantime, my PDX “friends'” tweets have opened my eyes to a whole new world of OpenID, dataportability, and the real value of social/business networking.
http://twitter.com/gwalter
@Gary: Looking over my geeky feeds, it looks like a few other failures can be attributed to Herr Jobs, Engadget, CrunchGear, FSJ, MacDailyNews, and more. I expect uncov to have some appropriate snarky comments by tomorrow.
It’s funny (or sad, depending on your viewpoint) to note: 1) what the Interweb cares about most and 2) how woefully unprepared these sites are.
@Gary: Looking over my geeky feeds, it looks like a few other failures can be attributed to Herr Jobs, Engadget, CrunchGear, FSJ, MacDailyNews, and more. I expect uncov to have some appropriate snarky comments by tomorrow.
It’s funny (or sad, depending on your viewpoint) to note: 1) what the Interweb cares about most and 2) how woefully unprepared these sites are.
Its indicative of how C-Ruby and Rails performs poorly, but not necessarily a reflection of Java-Ruby and Rails.
Most of the performance problems in Ruby/Rails (poorly written IO, poorly written thread management, poorly written garbage collector, poorly written memory management) can be mitigated by using what’s built-in to Java.
The JRuby guys have a chance to re-do Ruby the right way… they’re already getting better performance than optimized C-Ruby for number crunching, without even trying hard:
http://headius.blogspot.com/2007/11/java-6-port-for-os-x-tiger-and-leopard.html
Its indicative of how C-Ruby and Rails performs poorly, but not necessarily a reflection of Java-Ruby and Rails.
Most of the performance problems in Ruby/Rails (poorly written IO, poorly written thread management, poorly written garbage collector, poorly written memory management) can be mitigated by using what’s built-in to Java.
The JRuby guys have a chance to re-do Ruby the right way… they’re already getting better performance than optimized C-Ruby for number crunching, without even trying hard:
http://headius.blogspot.com/2007/11/java-6-port-for-os-x-tiger-and-leopard.html
@bex: I think Rich has been happy with the performance of Mix and JRuby generally so far.
I’ll let him comment more in detail on the larger issue of Rails vs. JRuby.
@bex: I think Rich has been happy with the performance of Mix and JRuby generally so far.
I’ll let him comment more in detail on the larger issue of Rails vs. JRuby.