4 USAGE and RETENTION/DELETION POLICIES

In any email system, clutter can be a major problem with some users better than others at keeping it under control. It can prematurely exhaust disk space, forcing administrators to upgrade the disks, crackdown on the laggards or both. An upgrade can mean the loss of a weekend but crackdowns often result in hair loss and little else. While most messages can be dismissed at a glace, a small percentage of then will need opening to check they are not important. It can take forever and a day.

There is always the brute force approach - kill all messages over a certain age, regardless of content. If users sebsequently realize they still need a particular message, there is always the archive - or so people can lull themselves into thinking. Unless the archive is properly indexed it is of little value. If a message proves elusive, people can develop a dogged determination to hunt it down. Given they may have remembered the message incorrectly, a poorly indexed archive can be a compelling invitation to waste time. And the archive must be readily accessible. If it consists of a collection of compact disks, access will be too slow. But if it is online, how is it different from the main email service?

There are no simple answers but there are simple approximations to answers. Most messages fall into obvious categories. Some are just babble, some are time dependent, many are just reminders, others will be about matters that are ongoing. And correspondents also fall into obvious categories. There will be personal contacts who babble and those who don't, data streams such as Google Alerts, companies that sell things of interest, people requesting price quotes or reporting problems, potential clients etc. Many of these can be categorized by their ingress. It is a reasonable assumption that incoming messages to jobs@myorg.com for example, are looking for a job. Correspondents also become categorized over time by the action of users and administrators. Dismissing an email as junk suggests the sender is prone to sending junk. It is all about searches. Search criteria aids categorization which in turn aids the formulation of search criteria, and it is the search criteria that drives and ultimately implements the retention/deletion policy.

Both automatic and user defined folders have numerous properties which control behavior. If these are not defined within a folder they are inherited from the parent folder (if one exists), failing this a default is applied. Among these properties are TTL (time to live), which auto-deletes on expiry. The TTL can be constrained by LCT (last conversation time), which keeps all the messages in a conversation until the most recent has expired. The folders relating to particularly important correspondents can be earmarked by the user or the administrators, either to expire after very long periods or not at all. Individual messages can be singled out for retention by the user, should they have noteworthy content.

Much however will be driven by users sending messages. ...