The Verdict's In: Ruby on Rails is Not Scalable. For Twitter, Anyway

Any hardcore Twitter user will tell you that for all its merits, Twitter's got reliability issues -- especially during periods of high demand, which isn't conducive to growth.

And apparently it's all Ruby's fault.

Following the departure of Blaine Cook and Lee Mighdoll, word has it Twitter will be abandoning Ruby on Rails.

TechCrunch says it'll be starting afresh with either PHP or Java.

When I attended the Ruby Conference last year, this was the big enthusiastic takeaway: "RoR is great for development!" But the lingering question was, "Is it scalable?"

Nobody had been on Rails long enough to know. Pro-Rails devs could only surmise. Most, including Blaine Cook, argued yes.

Michael Arrington writes:

Rails has always bred controversy. Developers have argued that it is fundamentally flawed and unscalable; others have argued back saying the opposite. Earlier this year, one of the core community members and creator of the popular Rails web server Mongrel abandoned rails and trashed the community.

I don't really know if the underlying architecture is Twitter's core problem or if there are other, more philosophical issues brewing back there. For the company, the immediate goal is to create a reasonable degree of stability for users. (Maybe then it can try making money.)

That won't necessarily be easy, considering Twitter's userbase is uniquely sensitive to site hiccups of seconds or minutes.

The good thing about switching to something like PHP is developers have more experience scaling it. So provided Twitter can transplant -- or slowly replace? -- two years' worth of infrastructure without a hitch, its long-term outlook looks less murky.

From a creative perspective, the departure is disappointing. In a way, Twitter is Ruby on Rails' first mainstream success. It's the true test of whether RoR is a legitimate architecture for enterprises, and not just geeks with discretionary time.

Still, the conservatism behind the approach makes sense. I like RoR and all, but if you're a company on a growth spurt whose top architect just exited stage left, this is a bad time to be playing Mad Scientist on a framework nobody really gets.