Gigaom (highly recommended for their frequent high quality posts on cloud computing) posts about the Dell/Wikia spat over upgrading hard drives in their servers.
The Wikia system is an excellent candidate for moving to the cloud. Wikia’s system is only 250 servers. There are much larger systems running on AWS. I’m sure they have the same kinds of concerns that we did at Melodeo when we first moved to AWS. We had to do a moderate amount of work on our solution to get it to work well. A common issue with a system transition is database utilization, as it is common for a self-built data center solution to feature a few very large servers for database (typically MySQL), and then a lot of the more cost efficient mid-range servers for components that are easy to distribute – software like Apache and memcache. In those days, the largest AWS instance was smaller than our big datacenter servers.
However, since the time we had to do some work to get our MySQL implementation optimized, AWS has introduced much larger instance sizes (with much better IO), and RDS.
Having to be worrying about hard drive upgrades in your racks of servers just isn’t as much fun as it used to be.
Rather than fight with Dell, why not offload that worry to someone else?
-bw