Cloud Migration Day

I started the week in search of an alternative hosting site for my open-source projects, now that Google Code is joining Google Reader in the steaming pile of great projects getting unceremoniously dumped because Google has never understood its user base. As I evaluated my options, I realized that it might be an opportune time to migrate my entire fiefdom into the Amazon Web Services cloud instead. Here are some of my justifications:

• It's hard to find a good, free Subversion host anymore. Though I've used Bitbucket and Git at work before, I find that the added complexity is unnecessary for my projects with a grand total of 1 developer(s).


• I love my web host of 7 years, Kattare, but it would also be nice to get more hands-on experience with AWS, now that I've got two AWS certificates generating money in the bank.


• I don't mind tinkering and rolling my own solutions. The URI! Zone was hosted on my college dorm room Pentium from 1996 to 2000.


• When things explode, I like the idea of being able to fix them on my own timeline. The 24-hour Bitbucket outage last month was a destroyer of productivity. (Ironically, my current web host went out for 2 hours right in the middle of my research for this post).


• The URI! Zone is not a mission-critical application, except during the hours when Doobie and Evil Mike are bored at work. Also, it has a very small audience, with daily visits in the double digits. Cloud failover and scalability are not a big deal to me, but they'd be nice to have in my back pocket if there were a sudden surge in popularity, due to my April plans to blog about nothing but Justin Bieber.


• Moving to the cloud would give me a lot of flexibility to quickly spin up new project sites (and tear them back down after I inevitably get bored with them -- see also, Auricle, Bugler, NoHunters, Team Turtle, Plinky's Tactics for Warsong Gulch, and Anna Had a Baby: Part I).


• It would also be nice to have a household backup strategy that didn't involve buying successively larger USB sticks every year.

Here's a systems diagram of the resultant cloud architecture that I'm envisioning:

Here's a cost comparison for my various requirements, showing how they are fulfilled today, how they would be set up in the initial cloud migration (assuming Reserved annual pricing), and how much costs might increase if I scaled up. My monthly cost would immediately increase from $20 to $30, which feels reasonable for the function and flexibility I would get.

The scaled cost ($60) is less palatable -- but the only reason I would need to bump up to that level would be for JIRA, which has been known to consume all available memory and small children in our work installations. I would probably consider outsourcing JIRA before I ever reached that level.

Have you had any experiences or horror stories adopting AWS? Would I be making a mistake by moving into the cloud? Share your thoughts in the comments section!