The JJW Blog
Research, Learn, Change
My running blog
Friendfeed
Archive
Regarding the specifics of your comments, the fundamental difference is on-demand. With AWS I can acquire or drop resources on a whim. At first, being able to just launch a bunch of servers or create 10 volumes from a snapshot “just like that” is a wow-experience and feels like a luxury, but after using this stuff for almost two years I just can’t go back. If you’re familiar with software development, when version control first became widely used it was a revelation to be able to tell a developer who was working on a special feature to “just branch the project and commit your changes to the branch”. Well, now we have gotten used to “just clone the staging system and test your changes there”, where the staging system is really a multi-server set-up and cloning that takes 5 minutes of clicking and editing a few web forms. We do this so routinely now that I can’t go back. Some scientific computing people we’re talking to are drooling over being able to create tens to hundreds of clones of a terabyte volume “just like that” because it allows them to get their job done quicker and better. Looking at Amazon’s storage volumes just feature by feature without considering the on-demand scaling is missing the big picture.
Amazon takes EC2 to the next level with persistent storage volumes « RightScale Blog
This sounds like the experience I am having on git. It changes my whole way of working with test changes. This AWS stuff is definitely pulling me in fast.
