home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Internet Info 1997 December
/
Internet_Info_CD-ROM_Walnut_Creek_December_1997.iso
/
ietf
/
pppext
/
pppext-minutes-95dec.txt
< prev
next >
Wrap
Text File
|
1996-06-03
|
8KB
|
203 lines
CURRENT MEETING REPORT
Minutes of the Point-to-Point Protocol Extensions (pppext)
Reported by Fred Baker, Cisco Systems
We are forwarding a variance request for CCP/ECP that indicates that
we will accept Motorola's letter of intent on these and advance these
documents. Frank has posted an Internet Draft of the variance to this
end. There will be an extended last call to consider this, going into
January. We should expect acceptance of the existing CCP/ECP
documents for Proposed Standard status in February.
Motorola's letter of intent states that Motorola will license the use of
its patents to all who apply for reasonable terms.
The problem in the variance proposal is the dependence on Motorola
carrying out its promise to reasonably license its patents. Vendors who
deal with Motorola should report their experience to the PPP list as a
test of this, as this report will be considered in taking the CCP and ECP
to Draft Standard.
IPCP Update to Draft Standard (Pall) draft-ietf-pppext-ipcp-
network-00.txt
Update to the IPCP left some ambiguity in RFC 1332. This draft
primarily clarifies RFC 1332 without adding new features.
Option 3 (IP address negotiation) is at issue.
Announcing an IP Address
Peer announces IP Address; you may:
o Ack the address
o Nak with a different (non-zero) address
o Reject the option (in effect accepting the address)
Peer requests 0.0.0.0; you may
o Nak with an address, assigning that address
o Reject the option (saying the link is unnumbered)
In addition, please add text to say that folks should not Ack this
packet.
Peer requests with no option; you may
o Say nothing in the Ack
o Nak 0.0.0.0; Ack with their own address or router ID. This permits
the system to ask its neighbor what its router ID is.
It was agreed that a note will be added pointing to other IP options,
such as the address of the DNS server. Those which may be negotiated
should use the DHCP INFORM option.
Updates to Draft Standard (Simpson)
o draft-ietf-pppext-chap-ds-00.txt
o draft-ietf-pppext-lqm-ds-00.txt
We are going to draft standard on these two, and these drafts are
clarification edits.
Implementors of these protocols are requested to send a note to the chair
identifying their CHAP implementations, and say with whom they
have tested interoperability.
US Robotics, Compatible Systems, Cisco, Bay Networks, Xyplex, and
3COM have done LQM implementations, but need an interoperability
test.
LCP Extensions
o Endpoint ID (Xyplex, Cisco, Network Systems): This is a go.
o Callback (Compatible, Cisco, Combinet, Digiboard): This needs
further updates for time before callback occurs. We also need to
discuss the option on the list
o Compound Frames: Let's drop this extension.
o Multiple FCS: Let's drop this extension
o Self Describing Pad: SGI?
Multilink: (Sklower)
o draft-ietf-pppext-des-multi-12.txt
There have been clarifying updates, requiring no operational changes.
Discussion of those edits will have to happen on the list, however.
Ascend MP+
There was a general consensus in the room that proposals made to
incorporate features from Ascend MP+ (which has advocates) should
not hold up revision and republishing of the Multilink Specification;
rather that these features should become additional control protocols
that might be useful on any dial line and not just a multilink dial line.
Multilink should target any line group, not specifically dial or ISDN
lines or any other specific line type.
Craig Richards (Shiva) will post a draft describing new control
protocols to implement the functionality of Ascend MP+. This will
differ from Ascend's proprietary implementation.
EAP Authentication: (Blunk et al)
o draft-ietf-pppext-eap-auth-01.txt
MD5 is the default authentication approach for EAP, and various
updates have been made per comments from Stockholm. One issue
raised there was the user interface definition in the specification,
which was too much. This was entirely removed in the current draft,
and that appears to be too much. The right level is a hint that says
that the password is a secret or something else, and therefore should or
should not be echoed. Token Card Type is debated; it may not be useful.
Network Express and Soliton Systems have implemented the RSA
implementation and interoperate.
Dave Carrel mentioned mutual authentication; there is a "man in the
middle" attack possible in CHAP, which would be obviated by one peer
requiring that the other peer also authenticate him. This is relevant to
Kerberos 5, which includes an IP Address in the credential. One
suggestion is to deal with this by using timing and challenges
subsequent to the authentication phase. Another suggestion is that
when A receives an LCP request from B which does not require
authentication and wants mutual authentication, it could LCP Nak
with an authentication option on the LCP Nak.
o draft-ietf-pppext-eaprsa-00.txt
Bill Whelan
RSA Type 9 Public Key Authentication proposal, using the ISO X.509
certificate.
Network Express and Soliton Systems have implemented the RSA
implementation and interoperate.
Certificate format appears to be problematic. The certificate needs a
type, preferably a well-known type, that is revocable and is available
to anyone with an RSA license and/or internet access. The issue here is
the purchase of ISO X.509.
Alternative Compression Proposal (Mader)
o draft-ietf-pppext-wcp-00.txt
Bay Networks Compression Protocol is a lightweight reliable
compression protocol. The issue is the perceived heaviness of LAPB.
Given the variance on CCP, it is not clear that this fills a real need, but
it has many good ideas. Keith will confer with Dave Rand on a follow-
on to CCP with the additional options. He continues to believe that
WCP itself should become a standard protocol, and is invited to make
that case on the list.
LZS-DCP Compression (Schneider, Friend)
o draft-ietf-pppext-lzs-dcp-01.txt
Add a bit to support moving the check field to the end of the packet, as
used by Stac's new hardware.
Option 23 padding - this should be done using self describing pad if
padding is required.
This was taken to the list.
PPP/FR and the Data Encapsulation Draft
A lunch meeting was held to discuss the future of these drafts, which
was attended by Keith Sklower, Keith Mader, Bill Simpson, Joel
Halpern, Andy Malis, Fred Baker, and Frank Kastenholtz.
We agreed that the basis of discussion should be a set of criteria. The
criteria we chose were:
Both drafts should be advanced as Proposed Standard if and only if
they had clearly differentiated applicability-if there were two or
more different problems and neither approach solved all of them.
Since the Frame Relay Forum has its own compression/encryption
specification, one or both should advance to Proposed Standard if and
only if additional PPP features are in view.
There has been an assertion that Data Encapsulation is the preferable
approach because it preserves the same data encapsulation.
Implementation experience is a guide to determining whether vendors
feel that this is a real issue.
We observed that three vendors, Xyplex and two unnamed vendors,
have implementations or implementations in progress of PPP/FR, and
no vendor has indicated plans to implement Data Encapsulation.
Therefore, we determined that preservation of the data encapsulation
is not a vendor concern.
We also noted that Authentication has always been mentioned as a
reason for parameter negotiation, and that some vendors already
implement Van Jacobsen Header Compression on Frame Relay.
Therefore, there is additional purpose.
We also note that all vendors present report market demand, and that
this demand specifically references PPP/FR. We could not identify a
set of applications where one would clearly prefer Data Encapsulation
given that PPP/FR was implemented.
Therefore, the consensus reached was that PPP/FR should advance
with all due haste to Proposed Standard, and that Joel and Keith
Mader would discuss the possible need for Data Encapsulation, and
would be free to publish it as an experimental RFC if they so choose.