Re: Reply to More on implications of 22.2

Richard Letts <R.J.Letts@salford.ac.uk> Wed, 06 July 1994 09:56 UTC

Received: from ietf.nri.reston.va.us by IETF.CNRI.Reston.VA.US id aa00967; 6 Jul 94 5:56 EDT
Received: from CNRI.RESTON.VA.US by IETF.CNRI.Reston.VA.US id aa00963; 6 Jul 94 5:56 EDT
Received: from mercury91.udev.cdc.com by CNRI.Reston.VA.US id aa02315; 6 Jul 94 5:56 EDT
Received: by mercury.udev.cdc.com; Wed, 6 Jul 94 04:55:58 -0500
X-From: R.J.Letts@salford.ac.uk Wed Jul 6 04:55 CDT 1994
Received: from zeus.cdc.com by mercury.udev.cdc.com; Wed, 6 Jul 94 04:55:53 -0500
Received: from io.salford.ac.uk by zeus.cdc.com; Wed, 6 Jul 94 04:55:44 -0500
Received: from salford.ac.uk by io.salford.ac.uk id <04693-0@io.salford.ac.uk>; Wed, 6 Jul 1994 08:36:56 +0100
Subject: Re: Reply to More on implications of 22.2
To: "Kevin E. Jordan" <Kevin.E.Jordan@cdc.com>
Date: Wed, 06 Jul 1994 08:36:50 +0100
Cc: S.Kille@isode.com, mhs-ds@mercury.udev.cdc.com
In-Reply-To: <2e1a22da0f76002@mercury.udev.cdc.com> from "Kevin "E." Jordan" at Jul 5, 94 10:19:53 pm
X-Mailer: ELM [version 2.4 PL23]
MIME-Version: 1.0
Content-Type: text/plain; charset="US-ASCII"
Content-Transfer-Encoding: 7bit
Content-Length: 2029
Sender: ietf-archive-request@IETF.CNRI.Reston.VA.US
From: Richard Letts <R.J.Letts@salford.ac.uk>
X-Orig-Sender: R.J.Letts@salford.ac.uk
Message-Id: <2e1a7fa201f4002@zeus.cdc.com>

Kevin "E." Jordan wrote....
> 
> > 1) Here, and in general, the spec aims to use the read operation to
> > the maximum extent.   This is because it is a reasonable assumption
> > that all DSA implementations in the context of any level of fanout
> > will optimimse the performance of the read operation.  
> 
> I understand the strong tendency to use only read operations.  In most cases,
> this is a good idea.  However, when this leads to proliferation of aliases,
> it is probably -not- a good idea and an alternative approach should be
> considered. 

Here is how we currently run PP (using tables)
if there is a match in the tables then delivery proceeds normally,
otherwise I make ise of the Administartion_asigned_alternate_recipient
to re-direct mail to a duaperl script. This performs a set of directory searches
to locate POSSIBLE recipients. It then returns a message to the originator
giving these suggestions, even if there is only one match and it is exact.

This is analogous to the situation in Kevin's MTA where if the read-operations
fails to find a recipient the MTA does a directory search.

In practice this imposes little load on our DSA because users normally only
mis-address (or under-specify) messages ONCE. After that they will use the
correct address. 

Looking forward I expect many more user-agents to be directory-aware in which
case the number of search-type operations on the DSA will increase purely
because that is what these interfaces do!
Unecessarily attempting to cut-down the number of searches performed is 
not worthwhile in the long-term as MTA searches will become minimal as MUA
searches increase!

performing delivery on searches is arguable [configurable ?]

Richard Letts  
-------------------------------------------------------------------------------
Network Manager                               mail:    R.J.Letts@salford.ac.uk  
University of Salford                         phone:     +44 61 745 5252
Great Britain                                 fax:       +44 61 745 5888