I'm waiting for someone to move up to the next level of abstraction and effectively merge the appengine and AWS approaches. Something like a web framework based on [insert-favorite-framework-here] but with explicit supports for SimpleDB, SQS, S3, etc. as well as callbacks for notifying the master container about load conditions so that autonomous or rule-based VM instantiation (and reaping) can occur in response to traffic patterns.
A concept like that could be merged with a virtual hosting idea to give kinda the best of both worlds - flexibility and hands-offness...
felix
· 1 year ago
That's a pretty interesting idea, I hope someone's working on it already. Another thing that would be most awesome is coming at the same idea from a different angle - if Amazon could just let EC2 instances grow arbitrarily gigantic, so you get one instance and it simply can use more and more compute cycles as necessary - or if it's a logistical effort, instead of adding more instances you add more "cpus" to a given instance. I suspect that would be workable for many but the top 1% of super cpu suckers like twitter - you'd end up with a massive instance but not mind bogglingly so.
All Comments
http://www.rightscale.com/
I'm waiting for someone to move up to the next level of abstraction and effectively merge the appengine and AWS approaches. Something like a web framework based on [insert-favorite-framework-here] but with explicit supports for SimpleDB, SQS, S3, etc. as well as callbacks for notifying the master container about load conditions so that autonomous or rule-based VM instantiation (and reaping) can occur in response to traffic patterns.
A concept like that could be merged with a virtual hosting idea to give kinda the best of both worlds - flexibility and hands-offness...