Thanks for the link correction, I’ve updated the post.

The ’super speedy’ hosting solution is just some linux machines running in a datacenter. I think it performs better more as a result of some kind of delays or rate-limiting that Twitter is imposing on the Google AppEngine infrastructure.

Although there are also issues with AppEngine (a) often taking a long time to apparently “reload” the app in some way every X number of hits to the site – not clear exactly what, but it seems to relate to the app being “swapped out” for lack of a better word. And (b) with URL-fetch operations, so any time an app uses external web services, such as REST APIs, it seems to suffer on AppEngine.

I would be reluctant to build anythign too serious on AppEngine, but it can be a good way to prototype and host smaller, less critical apps. If I prototype on AppEngine, I would be thinking from day one about how it moves to a stand-alone or alternative platform when it reaches a certain level of importance or scale.

What is your new ‘super speedy’ hosting solution? Does it have the scalability App Engine promises? It would be really useful to know as I’m making a decision about architecture for a project very soon.

Many thanks,
James