First up is the story about Google warehouse scale computing patterns. The idea that you’re going to design a data center for future applications is ridiculous when you have no idea what those applications are going to be. The key to this is a data center that’s flexible to the max and can change on a dime when requirements change, Build a data center for 5 year lifespan and you’re stuck with a forklift upgrade or abandon it and start over.
Second up is the salesforce.com cloud computing outage this morning. It’s not the cloud failure (it seems like all the major cloud have to play “me too” with those), it’s the idea that their stuff was in one basket enough to blow it all out of the water.