Re: SOS & LDBP Thu, 12 March 1992 17:22 UTC

Received: from by ietf.NRI.Reston.VA.US id aa00563; 12 Mar 92 12:22 EST
Received: from by NRI.Reston.VA.US id aa13873; 12 Mar 92 12:23 EST
Received: from by NRI.Reston.VA.US id aa13682; 12 Mar 92 12:23 EST
Received: from by with Internet SMTP id <>; Thu, 12 Mar 1992 13:22:41 +0000
Received: from localhost by (5.61/1.3-PSI/PSINet) id AA00272; Thu, 12 Mar 92 08:19:18 -0500
Message-Id: <>
To: Colin Robbins <>
Subject: Re: SOS & LDBP
In-Reply-To: Your message of Wed, 11 Mar 92 09:25:49 +0000. <"9981 Wed Mar 11 09:25:22 1992">
Date: Thu, 12 Mar 1992 08:19:15 -0500

> On its own, LDBP seems a reasonable thing to do (and I have even
> suggested some minor improvements to the authors!)  However the purpose
> of this mail is to ask if it is still needed in the light of SOS?

Thank you for bringing that up. When I first saw SOS, I asked myself
that question. Undoubtedly many others had the same thoughts too when
they saw the various documents.

In my opinion (and a very biased one it is too :-)), there is a place
for both LDBP-type protocols and the SOS.

LDBP, and protocols like it, are more oriented toward application-level
functionality. A close look at LDBP will reveal that it has
minimal presentation/session functionality, and what functionality
there is in those areas has been integrated into the protocol itself
(ie., they are not distinct from the application protocol itself).

In contrast, the primary thrust of SOS is the definition of an
integrated ROS/ACSE/presentation/session stack that will offer
OSI-like services to application protocols. Although three of the
appendices in the SOS I-D define application protocols, that is
not the main theme of the SOS.

So, in some sense the two are "apples and oranges".

> SOS is a much more general than LDBP, and describes a more general
> lightweight approach for the full OSI stack.  Appendices A and B of
> this document describe two example SOS nameservices protocols, based
> upon X.500, and something similar the LDBP could easily envisaged.

Exactly. SOS is not an LDBP replacement. It is the stack that would
provide lower-layer services to an LDBP-like application protocol
that would sit on top of that stack.

> when [SOS]  arrives, there is a possibility, it will
> make LDBP redundant.

While this is possible, and only time will tell if anything becomes
redundant, I believe there is a niche for both LDBP and its ilk,
as well as the SOS and the protocols that will use its services.

LDBP and the SOS both represent tradeoffs in OSI deployment.
At one extreme is the full-OSI stack in all its full blown
generality with accompanying complexity. At the other extreme is
things like LDBP which has OSI-like application functionality
but a minimum of the underlying stack: the tradeoff was decided
in terms of specificity (only what was necessary for that specific
application was taken from the OSI lower-layers), with accompanying
reductions in complexity and hopefully smaller implementations.
The SOS takes a middle ground: it defines an
application/presentation/session subset useful to a subset of
application protocols (although that subset does turn out to include
most OSI applications), trading off some simplicity in favor
of increased generality.

So, they could be viewed as complementary to each other, not
competitors for one another. Depending on resource and other
restrictions, implementors and users will have to decide on
their own set of tradeoffs: now they have three choices (LDBP,
the SOS and the full OSI stack), instead of one.