Twitter Back On The Straight And Narrow: Interview With Founder Biz Stone
Duncan Riley
24 comments »
On May 17 I shared my frustrations with constant downtime on group instant messaging service Twitter. The responses were interesting. Twitter users are passionate about the service and many see it as a mission critical application.
A lot can change in the space of 2 weeks. Since writing that post Twitter’s uptime has improved remarkably. There are still downtime issues; Twitter was offline for over 5 hours Monday according to Pingdom; however these figures are deceptive as there is a noticeable improvement in quality when Twitter is up. Twitter has not been plagued by intermittent functionality failures such as API support issues that block third party tools such as Twitbin from connecting, as was the case two weeks ago.
I asked Twitter’s Biz Stone what was going on behind the scenes at Twitter.
Has Twitter become more stable following the issues earlier in the month? What has the Twitter team learned from the experience?
Biz Stone: “We’ve worked to isolate various parts of our system to prevent one outage from taking out the entire network. Investigation into the causes of previous bouts of instability have taught us a great deal about our usage patterns and shown us a clearer path towards being prepared for future waves of growth and usage”.
In what way has Twitter’s reliability improved?
Biz Stone: “Our recent efforts towards scalability allow us to stay on our current growth curve, and even increase the rate as we add new ways to send and receive updates such as our recent Facebook application”.
What next in terms of scalability?
Biz Stone: “We’ll continue to isolate different aspects of the system and perform micro-optimizations of heavy-usage areas. We’re currently implementing additional internal tools which will enable us to do more meaningful research of usage patterns. Finally, we are adding more machines”.
Let’s hope Twitter stays on the straight and narrow.
More crunch: click here for the Twitter fact sheet.
Sounds good. I am curious to learn a bit more about the behind the scenes challenges with Twitter’s up time. Probably much to learn from. Duncan - A longer interview maybe?
Yea, also, Twitter is taking on users like it’s the cure to some major disease. Are the problems going to come back once they scale up again?
I’m stilling wetting my feet in the Twitter universe (I like Jaiku a wee bit more), but if they can keep the product stable I think they could capture even more of the marketplace than they already have.
And that would be saying something.
Wow, that’s some insightful interview. Maybe next time you can give them a phone call and ask some follow-up questions.
Wow - to think that there are a good number of people who consider Twitter a mission critical application…wild.
Even as a long-time web developer and gadget/technology freak, I’m still trying to figure out what purpose (if any) Twitter actually serves. Congrats to them for the uptime improvement though. I’m looking forward to seeing how the service progresses.
just curious about their backend (ruby, .net, lamp?) … I like reading about companies scaling from start-up to meet the demand of users. I will search for the url, but there’s a neat blog entry on digg’s scaling.
And what again is Twitter’s business model? How do they generate revenue?
Are they getting a cut from telco for text messaging fees or simply hoping to be acquired?
@phugit (#5)
They use Ruby on Rails, I think it might even be considered the largest Rails app on the web. Here is an interview with Alex Payne that talks about scaling issues with Twitter.
And also, here is a link to the top 10 presentations about scaling web apps. Hope that satisfies some curiousity…-Metagg
@Jake
I have absolutely no idea but wonder the same thing.
I like twitter and respect what they’ve done. Who’da thunk what kind of crazy traffic and demand they would see so fast - it’s been amazing. Hopefully they’ll be able to stay strong enough to keep up, there have been a few windows of opportunity for other systems (Jaiku, as noted in the comments) to move and they have moved fast… but Twitter can still pull this off - give good core service, improve the uptime and speed, and keep open (for other development/API opps). The interface with Facebook is cool, who knows how valuable… anyway, my hat off to what they’ve done.
Jason Alba
CEO - JibberJobber.com
If you twitter yourself out of a job you’ll need JibberJobber :p
phugit, you can figure it out from their job descriptions
Ruby on Rails Developer
* Comfortable working-in/learning-quickly a Ruby on Rails environment.
* A craftsmanship-like approach to writing code.
* A minimalist design aesthetic with an eye towards reduction.
* Ability to work on a small team and get lots done with minimal oversight.
* An evident passion for the intersection of the web and mobile technology.
* Deep knowledge of your chosen tools and how they interrelate, which should include at least one scripting language (Ruby, Python, JavaScript), one markup/style language (CSS, HTML), and one application-specific language (C, SQL, Objective-C).
* Ability to articulate ideas and future direction to both the rest of the team and users.
Senior Engineer
* BS or MS in Computer Science or equivalent experience.
* 5+ years of real-world software development experience.
* Experience architecting high-volume web applications and/or messaging systems.
* Excellent and influential communication skills with engineers and non-engineers.
* Extensive experience programming in both scripting and application-specific languages.
* Deep familiarity with Unix environments, HTTP, TCP-IP.
* Experience with Jabber, Ruby on Rails, and MySQL a plus
* Strong interest in Twitter and developing a world class Internet application
They were down 5 hours on Monday (2 days) ago, I think it’s a little soon to be touting new reliability efforts
Many people moved to Jaiku recently, at least in Germany. There were several outtages - not only on Monday.
Ping from Twittown - Twitter Community and forums
http://www.twittown.com
Did Biz say anything specific to Twitter? Reads like a generic marketing script for an unreliable service. Maybe he cribbed it from an old Cingular or Comcast press release?
Great find! Thanks for posting this, that’s my something new for today learnt.
———————————
Free Download
http://www.onedownload.org
5 hours of down time is something else and DELETING all of my messeges on twitter is totally something else!
Now a happy jaiku addict!
Normally Techcrunch loves to point out when sites are powered by Ruby as if it really makes much difference at all (part from the fact they’ll just have hiring problems).
Well in this case, clearly it has for all the wrong reasons. So let’s hear it TC! Twitter is a Ruby site and has scalability problems!
Ruby has scalability problems!
Let’s hear your Ruby fanboisms next time…
Not a lot of improvement these days…
IM notification is wild… getting on and off all along the day…
Then when messages sent to the Twitter buddy are taken into account (which is rarely the case) they get mixed up with other messages finally sent from the web as the web is more stable, making the whole thread completely messy…
Looking for a serious alternative now…
Hello! Good Site! Thanks you! kcjxcstqfz