just throw more servers at it, they said….
When I was looking into Ruby on Rails, I was always a little sceptical about the whole “just throw some more servers at the problem” answer. Sure, servers are cheap, but once you get to internet-scale, they get quite expensive, quickly – becuase you need a LOT of them. And now this, from Twitter and El Reg:
Today, Payne said, most of the hip dev set codes in Ruby or PHP or Python because they’re perceived as “agile” languages suited to quick site changes – but also because the deverati grew bored with languages like Java and C++. The trouble, he insisted, is that the so-called Web 2.0 languages aren’t always as efficient as they need to be, especially in an economic environment threatening to bring Web 2.0 to its knees.
“Investors now want to know that you’re not going to be paying tons and tons of money for servers because you decided to build in a stack that isn’t as efficient as possible,” he said.
“In a down market, you may have to make choices between technologies you love to work with and technologies that will keep the lights on. The question is… whether we can afford to build things that we love with tools that we love. The answer is ‘Yes’. But it means adopting a different set of tools.”
Bolding is mine, natch. For smaller sites, there really is no difference between the various languages and platforms. But once you get to, say, 500 servers, it starts adding up quickly (esp if you dont have a paid-for business model!).
The question, especially given the current financial climate, as changed. It was
Can you scale Rails*? Yeah, of course we can scale Rails – just design it in.
To
Can you AFFORD to scale Rails*? How many boxes do you need to service 1 million users? How much power? Disk? Bandwidth? etc?
*replace Rails for PHP, Python, Grails, Java, .NET, whatever else, as needed.
BTW, I’m aware you still need to design things to scale. But using something which requires a scale out to work at modest load is rapidly becoming a non-starter. I think things like EC2 and GAE are going to take off soon – if only because they can get enconomies of scale that others can’t.
I’m sure the US (and UK) will find out when bandwidth is no longer free (or rather “unlimited”), a lot of things on the internet simply dont work. Online backup is a big one – if you have a 100GB cap (or 20GB which is common in New Zealand), do you want to back ALL that data up into the cloud? Would you maybe get a better result – and cheaper – by buying a 500GB disk and storing it at work?
