I think it is a good idea to revisit the caching in subman to maybe move to python-rhsm. The caching behavior seems to be on network bound operations. Furthermore it seems to have the most utility when the connection to the server is broken in someway (that is when most of the caches are used). Its almost like we are testing the connection object that is passed into the cache objects and returning the contents of said cache object if the connection object fails the test (the update from/to the server).

The suggestion is to move the caching to be part of python-rhsm. I would also like to add that it would be very nice for these modifications to include a method of determining the source of the returned value for all cached resources (basically it would be neat to have a method that would return the source of the last result).

These improvements could include a separate connection object that caches all interactions with the server but does not necessarily need to.