While a number of people are attempting to rewrite the mail tool and mail service from scratch, I propose that we make some simple modifications to the existing tool to alleviate the performance problems. This will be done at a cost of reduced functionality when there is a large number of messages - but will be upwards compatible with the current data models and applicable as a patch.
If you want to argue that the entire data model needs to be redone and the entire tool needs to be redone - that is fine. You can do that in another JIRA. That will take too long.
We have a problem that needs solving now so that large mail archives do not compromise our servers. This will not be done in time for 2.5 - but it can go in the maintenance branch/release