Comments on: Evolving Services on EC2

By: HappyCodr | Evolution

HappyCodr | Evolution — Sun, 11 May 2008 22:23:41 +0000

[...] Compute Cloud over the course of 1 year, written by Geezeo’s star system administrator. elwoodicious.com Tags: amazon, ec2, [...]

By: james

james — Sat, 10 May 2008 16:05:24 +0000

Well, it was really the interseection of price and performance. Anyway you build it a redundant cluster is pricey much more than replication and not to mention back ups and recovery is greatly complicated.

For grabbing ips, that can easily be done with a shell script wrapper around the run command or one that runs on start. Up and checks it in either to a homegrown service or just writes to a file on another instance. One of the things I’ve been meaning to get to is making elastic slaves that will come up, grab the latest backup and master info, source in, register, and start replicating. Maybe I’ll move that up in my To-Do.

By: Clay Fouts

Clay Fouts — Fri, 09 May 2008 22:33:30 +0000

The ability to cheaply experiment with different designs is one of the things I like most about using EC2.

Was performance of the small VMs the only reason you decided not to go with MySQL clustering? I’m thinking about implementing that as the DB for part of our service and am concerned about difficulties arising from some of EC2′s idiosyncrasies, like never knowing what IP an instance will get and so forth.

By: Katie McCaskey

Katie McCaskey — Mon, 05 May 2008 14:06:53 +0000

I love the diagrams! I know for sure that the site works a lot faster thanks to all your work behind-the-scenes. What would we do without you?

By: james

james — Mon, 05 May 2008 00:45:13 +0000

Thanks! Awesome team to work with making the best product out there!

By: Peter Glyman

Peter Glyman — Sat, 03 May 2008 11:32:36 +0000

Thanks for building such killer architecture James….you da man!

Dugg: http://digg.com/tech_news/Evolving_Services_on_Amazon_Web_Services_AWS_EC2

Pete