On 6/5/06, Kara Rubinson <[log in to unmask]> wrote:
> Has anyone successfully implemented an email archiving solution?  If so,
> I would love to hear your thoughts on the following...

As mentioned by others, and you say you're already aware, what you're
looking for is being misrepresented by IT as an "email archiving" product.
And while YOU MAY UNDERSTAND THIS, it's important that you make sure IT
understands this, because they may think the only concern is to "archive the
email", NOT TO MANAGE IT.  It's a major chasm to cross and if it isn't made
clear early on, they may be looking for something completely different than
what's needed to achieve the goals.

I understand & agree with the need to retain/dispose of email based on
> content & the retention schedule.

I think with you being in a public school system, and also in a State
environment, you have other concerns beyond those of many in private
industry... simple retention based on content.  You have two other
considerations that are actually a double-edged sword... privacy and public
records laws.  A critical component in an e-mail management system for your
environment may be policy and training to ensure users know what should and
shouldn't be expressed in e-mail; and what repositories need to be isolated
when the content is backed up, etc.

Does your system require email users
> to "declare" emails as records?

Most systems will require this.  You can do a certain amount of "rules
based retention model" development, such as all messages from the
Superintendent to a certain level of Administrators might be declared a
record automatically, but the detailed retention period for these would
still be based on the actual content.  So, although it may be declared "as a
record", it would need to be examined further to determine the retention
period of the record series it belongs to.

Or, does your system automatically
> archive emails for different retention periods based on keywords &/or
> the To/From?

One of the problems with keyword classification is you need to index the
entire content of the e-mail and then build a system that searches for a
combination of words within the body of the message to determine a
contextual "value" of the message and then classify it.  And as mentioned by
someone else, we're not there yet. You can build it over time, verify and
validate the level of  success, then tweak it continually until you reach a
threshold you're satisfied with.  Even then, you may only get 75-80%, and if
you have 20,000 people, with 100 messages a day, that's 2 million
messages... and 20% that would need to be reviewed is STILL 400,000 messages
a day. If there's only 10 messages a day, it's STILL 40,000.

Or some combination of both?

You run the risk of saving way too many messages for way too long, or way
too few messages for way to short a time. And that's one thing to take into
consideration... RISK.  What is the risk associated with mis-classifying an
e-mail record?  Is it not being able to find it, or having it around too
long and it being a potential ticking time-bomb, or the wrong person finding
it and seeing information they shouldn't be able to see? And naturally, the
cost of managing it.

With over 20,000 email users
> of widely varying technical abilities, we hope to make our system as
> automated as possible, keeping user involvement to a minimum.

A simple "3 bucket classification system", one that calls upon the user to
identify the e-mail as either:

1) Non-record material, (personal or of no administrative value)- discard
2) Record (transitory or strictly administrative in content)- discard within
2 years
3) Record (requiring retention determination based on content)- period
varies, but 2 years plus

May be one way of starting out.  And one way to implement it might be to
have a pop-up window open before closing the message, requiring the user to
make this determination.

Is that do-able?

EVERYTHING is do-able... how much time and money do you want to devote to it
is the question.

Larry Medina
Danville, CA
RIM Professional since 1972

List archives at
Contact [log in to unmask] for assistance