CURRENT_MEETING_REPORT_
Reported by John Veizades/Apple
APPLEIP Minutes
The Working Group met and discussed work in the following areas:
SNMP
Work in the SNMP area is split into three areas. The AppleTalk MIB Plus
(the first version is now RFC 1243) (this MIB will no longer be called
MIB 2) is now out for comment as an internet draft. There are
implementations of the RFC 1243 MIB available on shiva, cayman,
farallon, 3com and acc systems. Implementation and use experience has
led to the following list of problems with this MIB: it is felt that
there may be more variables than is needed, this MIB does not allow for
the configuration of routers and there are questions on if this MIB
supports half routers well. It is felt that there are significant areas
for discussion and implementation the group is not trying to rush the
MIB Plus document and is waiting for appropriate comment. The SNMP over
AppleTalk document is ready to move forth as a Proposed Standard and
will be doing so shortly after comments from this meeting are
incorporated in the document. Concern was raised about getting major
console manufactures to incorporate this standard into their consoles.
Concern was also raised as to the ability of the MIB to be used for the
global changing of a network's zone list. Test tools are available from
Mike Ritter (MWRitter@Applelink.apple.com). The last item was The
Macintosh system MIB this is now out for general comment.
AURP
The AURP (Apple Update Based Routing Protocol) will be progressing from
internet draft to proposed standard after revising the state diagram.
The completed document will be submitted as a Proposed Standard in the
Internet community as well as being available as an APDA document. A
vendor product bakeoff is scheduled for MacWorld in January, there are
about seven companies at various stages of implementation (cayman,
cisco, shiva, dec, farallon, compatible systems, novell, 3com, pacer and
Apple). Seeding of some of these products to sites around the world is
also planned in the next few months.
ABGP
A presentation was made on the possibility of introducing a BGP like
protocol as a border gateway protocol for AppleTalk. Greg Bruell from
Shiva made the presentation, Yakov Rekhter (IBM) and Scott Brim
(cornell) were in attendance. Why bgp? It looks a lot like aurp when
you make some needed extensions to BGP to incorporate AppleTalk.
Transport stays the same as BGP except that it uses a different tcp
port. Message layer stays the same as BGP, the autonomous system number
maps to the domain identifier and there is a change to the network list
into the network zone tuple list. Some advantages are that BGP is a
1
well known implementation, decisions on policy are outside of the update
protocol and it is easy to implement.
AppleTalk and OSPF
Greg Bruell from Shiva led a discussion on using an OSPF like protocol
to replace the AppleTalk IGP which is RIP like.
PPP and AppleTalk
The document presented is close. Comments will be incorporated and
reissued as an internet draft for comments from the appletalk community
as well as the PPP community. Additions to the current document include
calling out and describing the operation of several common cases; node
to server, node to node and half routing. Comments on hop count
incrementing and which options should be negotiated for each case will
be added. Operation with AURP will be left to the AURP effort. The
smartbuffering compression algorithms are available through APDA in the
document which describes the operation of the Appletalk Remote Access
Protocol (ARAP). Implementations are in progress by cisco, cayman,
shiva, novell, telebit, A/UX and Farallon. The AppleTalk over PPP work
was presented to the PPP Extensions Working Group. The PPP Extensions
Working Group added functionality that will allow all that is needed for
call back in the security fields of the LCP. Both brad Parker and John
veizades presented the apple communities view on dial back and security.
The version of the PPP document that will contain the PPP extensions for
security will include everything needed for dial back as presented in
the ARAP specification as well as the ability for the user to specify
the number string to be called back at. The security specification will
also contain whatever is necessary for ``secure ID'' extensions.
MacIP
Three outstanding comments were brought up, they will be incorporated
into the current document and it will be posted for final review before
moving the protocol to proposed standard. The areas of comments were
ICMP messages, out of zone operation and multiple servers in the same
zone. In the area of ICMP messages it was decided that ICMP redirects
will be gleaned by the macIP gateway when it is doing proxy arp for
nodes in the AppleTalk network that are on the same logical subnet as
the gateway. In the area of out of zone operation if two host use the
same address in the AppleTalk internet packets destined to one will be
reliably dropped. When two servers are in the same zone some election
mechanism will be used to choose one of them as the gateway though
others will be kepted to use as secondaries if the first fails to
provide registration or services. Two features should be added for the
rebuilding of the AppleTalk address to IP address mapping on server
restart one is the Phil Koch algorithms for gleaning address mappings
and the other is the ability to send NBP lookups to specified zones to
rebuild the mapping table.
OAF - Open AppleTalk Federation
Discussion were held on how to continue the growth of the infrastructure
2
related AppleTalk protocols. Most ideas evolve them by moving them into
the IETF community. Work is being done on charter definition, vendors
buy in and on discussing these issues with the relevant Apple people.
This effort would proceed within the infrastructure of the IETF, the
IETF has been approached as to the viability of this undertaking and
they advise that the work could be accomplished under an AppleTalk
directorate within the IETF. Concern was raised as to Apple's role in
such a venture and what Apple's commitment to such a venture would be.
Attendees
James Beers beers@nr-tech.cit.cornell.edu
Craig Brenner Brenner2@applelink.apple.com
Gregory Bruell gob@shiva.com
Philip Budne phil@shiva.com
Peter Caswell pfc%pacvax@uunet.uu.net
Richard Cherry rcherry@wc.novell.com
Richard Cogger rhx@cornellc.cit.cornell.edu
Peter DiCamillo cmsmaint@brownvm.brown.edu
Dino Farinacci dino@cisco.com
Karen Frisa karen.frisa@andrew.cmu.edu
John Gawf gawf@compatible.com
Bob Jeckell robert_jeckell@nso.3com.com
Holly Knight holly@apple.com
Louise Laier laierl@applelink.apple.com
Joshua Littlefield josh@cayman.com
Greg Merrell merrell@greg.enet.dec.com
Greg Minshall minshall@wc.novell.com
Robert Morgan morgan@jessica.stanford.edu
Michael Newell mnewell@nhqvax.hg.nasa.gov
Chandy Nilakantan csn@3com.com
Alan Oppenheimer oppenheimer1@applelink.apple.com
J. Bradford Parker brad@cayman.com
Christopher Ranch cranch@novell.com
Michael Ritter mwritter@applelink.apple.com
Eric Smith
Evan Solley solley@applelink.apple.com
David S.A. Stine dstine@cisco.com
John Veizades veizades@apple.com
Lee Wade wade@nsipo.arc.nasa.gov
3