When working with GAE datastore you want to execute minimum number of queries necessary to avoid delays and limit the use of DB API calls. Fetching comments one by one is not the best option then (even if done with ajax).
Probably the most efficient way would be to store the list of comment ids as an attribute of each article entity. Then you could get all comment ids with a single query. Having the ids you could build a list of GAE keys, and fetch all needed comment entities at once. We discussed that recently here: https://groups.google.com/d/topic/web2py/7dvo_jqIR38/discussion I'm also guessing that this might be quite common GAE use case and you might find a better solution already optimised for this on GAE forum. The other option is to show comments only on the article individual page (not on the list of articles). If you use the wrapper for caching, then you could memorize the entire response, so this is very efficient as the views are not rendered but read straight from cache. If you need some extra parametrisation inside the controller you could either move some of the code to a separate function which output is cached or simply cache only the queries: http://web2py.com/books/default/chapter/29/6#Caching-selects
