Conference 7.286::fddi

51.0. "IBM Will Rebuild Source Routing" by DELNI::CALLAHAN () Tue Apr 10 1990 10:22

TITLE: 

IBM Will Rebuild Source Routing: Admits the Market has Spoken

SUMMARY: 

IBM 802.5 committee representatives have proposed scrapping current
draft standard for Source Routing.  A new version will be written from
scratch and will establish Transparent Bridging as the base-level for
802.5 bridging, with Source Routing as an option. 

IBM acknowledges that it is not possible to legislate what goes on in
the market.  When it comes to the Transparent or Source  Routing
bridges the market has spoken.  Transparent Bridging is the standard,
open technology.  Source Routing is a proprietary non-standard
solution.

DETAILS:

IBM presented a proposal at a recent joint meeting of IEEE
802.1/802.5, March-1990 in California, which detailed major changes to
the current draft standard for Source Routing.

IBM proposed scrapping the existing Source Routing standard and
rewriting the standard from scratch.  The new standard would establish
all bridges as Transparent Bridges as defined by 802.1d and give a
special bridge, called an "SRT" bridge the option of handling Source
Routing packets.  In other words, the SRT bridge would be a
Transparent Bridge with the option to do Source Routing as well.
Essentially the proposal implies that from this point forward there
will be no more Source Routing-only bridges in the standard.

Prior to this proposal, Transparent Bridges as defined by IEEE 802.1d
discarded all Source Routing packets and Source Routing bridges as
defined by the IEEE 802.5 discarded all Transparent Bridge packets
(packets with no Source Routing info in them).  This made
interoperability between the two bridging domains virtually
impossible.  The 802.5 committee struggled with the interoperability
issue (dubbed SR-TB Appendix Z) for many months and came up with
proposal for an SR-TB bridge which was unworkable. 

Even so, an IBM group in Rochester, Minnesota outside of the
architectural control of IBM's Research Triangle Park in Raleigh, NC, 
went so far as to create a product based on the SR-TB Appendix Z. The
product was called the IBM 8209 LAN Bridge and required homogeneous
domains on both sides (TB for Ethernet and SR for Token Ring).  The
8209 only accepted SR packets on the Token Ring side and TB packets on
the 802.3/Ethernet side.

       /~~~~~~~~~~~~~~~~\             /~~~~~~~~~~~~~~~\
       | SR-only Domain | ----8209----|TB-only Domain |
       \________________/             \_______________/

Because of the proposal accepted at the joint 802.1/802.5 meeting,
Source Routing-only bridges now have a limited life-span.  As a
result, the IBM 8209 LAN Bridge will have a limited use in commercial
environments and will certainly not have the "SR side" pointed at FDDI
in the future.  In addition, it is now unlikely that Source

		       * DIGITAL CONFIDENTIAL *
    
Routing-only bridges will appear AT ALL on FDDI.  This means that FDDI
LANs will NOT become part of "homogeneous" Source Routing domains, and
thus our strategy of using Transparent Bridging as a means to
integrate our installed base of 802.3/Ethernet with FDDI is further
solidified.

IBM gave several reasons at the joint meeting for making such a 
proposal:

1) Transparent Bridging as defined by 802.1d is a widely accepted 
standard.


2) Requirements for Source Routing are only coming from some of the 
market.

3) The existing Source Routing draft standard (including the SR-TB
Appendix Z) lacked harmony with Transparent Bridging. 

 - Made for difficult interconnection of homogeneous domains (SR-TB)
   and required too many rules about domain interconnection.

 - Caused interoperability problems and degraded network operation 
   in heterogeneous domains.

 - Confused customers.

4) Customers are asking for the freedom to mix both SR and TB stations
and SR and TB bridges on the same LAN. 


The Digital Point of View
~~~~~~~~~~~~~~~~~~~~~~~~~
This is good news for Digital for a number of reasons.  We have been
telling customers that we committed to supporting standards and that
Transparent Bridging as defined by IEEE 802.1d is the standard bridge
algorithm, with the greatest potential to become the true multivendor
solution.  This recent announcement by IBM proves that we were correct
in our assessment.  

Customers who invested in Digital's bridge technology will see their
investment preserved, even enhanced by the emergence of next
generation LANs like FDDI, where those who invested in Source Routing
will have to make use of proprietary gateways to connect their now
non-standard, proprietary Source Routing "solutions" into the next
generation of technology.  Digital's installed base of bridges is
transparent to the applications that run across them and will extend
those applications to the FDDI environment.

From the extent of the changes proposed to the Source Routing
standard, it appears that IBM's product implementation has not been
without its problems.  It should be clear to customers now that Source
Routing would not have been appropriate for large-scale multivendor
backbone environments like FDDI.  The market has chose Transparent
Bridging because it is a proven algorithm and a standard.  Transparent
Bridging is literally the only existing standard technology for
bridging, and the only credible multivendor solution.

		       * DIGITAL CONFIDENTIAL *




		    Digital Party Line on This Issue
		 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

1.  Digital supports this move by IBM.  We have always believed that
    Transparent Bridging provides the most stable, scalable and
    manageable extended LAN.


2.  Digital believes that Transparent Bridging is the true multivendor
    solution and the marketplace bears this out. You cannot legislate
    what goes on in the marketplace and the market has spoken. 
    Transparent Bridging is the most widely accepted standard for
    bridging in the industry today.  


3.  Digital is committed to working in the standards process to make
    it easier to integrate all IEEE 802 style LANs with other
    technologies such as FDDI.  Since all bridges are now Transparent
    Bridges, we feel this goal is that much closer to becoming a
    reality.


4.  Digital believes that Transparent Bridging is the bridging
    algorithm of choice for FDDI.  FDDI is a multivendor LAN standard. 
    Transparent Bridging is the multivendor bridging solution.  In
    fact, multiple vendors have announced FDDI bridge products based
    on the 802.1d Transparent Bridging standard.

    This seems to be excellent news. If I read this right then it means .3
    and .5 networks will be able to interoperate easily via FDDI backbones.
    Is this correct or are there other complications??
    
    Somebody help as a civil war is presently taking place in our office
    about this subject!
    
    Bob Billers

    in the area of bridging in fddi what concerned me was that the transparent
scheme required more computational power in the bridge. 

digital for a long time was the only vendor that was able to make high 
performance 10/10 filtering bridges and that was using a sophisticated 
multiprocessor design in the bridge. 

now looking at, let say a 100/100 bridge. i wonder if it is possible to build
one that completely keep up. and what about higher speed lans.
with a source routing approach this would be simpler.

No, source routing is harder than transparent bridging, since (even before
the events described in .0) source routing bridges had to do SOME things
in the transparent bridge fashion.

100/100 bridging at speed is definitely doable.  I've already seen one
vendor's announcement claiming that they do, though that's an unverified
claim.

	paul