Gotta love sensationalist headlines...<p>Should read: Facebook CTO says purchasing servers for a separate startup prior to Facebook <i>might</i> have been unnecessary<p>The idea that Facebook wouldn't need its own data centers is pretty ridiculous, and that's what the title implies<p>EDIT: looks like "while at FriendFeed" was added for clarity ... Further edit: The title is changed now, but the original said "Facebook CTO says purchasing servers was his biggest mistake" or something of the like
friendfeed was built in 08, before amazon had released EBS as a feature. there is no way in hell the instance-store disk i/o would have kept up with friendfeeds mysql demands (you may remember the post Bret himself wrote on their nifty approach to mysql: <a href="http://bret.appspot.com/entry/how-friendfeed-uses-mysql" rel="nofollow">http://bret.appspot.com/entry/how-friendfeed-uses-mysql</a> ) Frankly, even EBS probably wouldn't have. Plus, back in 08 the larger ec2 instances were still pretty low on ram compared to what you could cheaply cram in a colo'd server, putting even more pressure on the disk i/o.<p>I think Bret's really wishing what almost everyone is, that these problems would just go away and you could pay a reasonable premium for that, but we're not there yet and certainly weren't in 2008. I don't think he really checked his numbers on this one.<p>edit: however there is a happy medium. managed service providers will gladly charge you ec2 prices (250/month for a low end server, 500/month for a beefy one) while handling all midnight colo trips for you.
In other news, Amazon's CTO says his worst decision was to buy servers as they could have just rolled EC2's API out on Dreamhost's Private Servers offering..<p>Less facetiously, this "regret" surprises me. I'd have though that of anyone, Facebook would want to have full control and security over their own servers. Agreeing to a contract where Amazon can pull the plug over "objectionable" content or even adverse effects on their network poses a significant risk for any large business to accept.
Until some very smart people get together and restart tandem, buying hardware will be a poor decision. Designing for a fail-by-design architecture will be the right thing to do. and if you have fail by design, why not just ship the executable to whoever's the cheapest virtual hosting at the moment.<p>PS: To very smart people: Please get together and start Tandem: the sequel. Pretty, pretty please. With sugar on top.
The <i>only</i> two paragraphs that actually match the headline:<p><i>If we had to guess, it's probably going to be orientated around media or news, because they are so social. When you watch a television show with your friend, it's such an engaging social activity.<p>We think that there's a next generation of startups that are developing social versions of these applications, where what Zynga is to gaming, they will be to media and news, and we're really excited about that.</i>