IP over ATM requirements: IP over ATM Working Group's Recommendations for the ATM Forum's Multiprotocol BOF Version 1
IETF recommendations for the ATM Forum's work on connecting IP networks with ATM.
Laubach, M., "IP over ATM Working Group's Recommendations for the ATM Forum's Multiprotocol BOF Version 1", RFC 1754, DOI 10.17487/RFC1754, January 1995, <https://www.rfc-editor.org/info/rfc1754>.
@misc{rfc1754,
series = {Request for Comments},
number = 1754,
howpublished = {RFC 1754},
publisher = {RFC Editor},
doi = {10.17487/RFC1754},
url = {https://www.rfc-editor.org/info/rfc1754},
author = {M. Laubach},
title = {{IP over ATM Working Group's Recommendations for the ATM Forum's Multiprotocol BOF Version 1}},
pagetotal = 7,
year = 1995,
month = jan,
} RFC text
Contents
Network Working Group M. Laubach
Request for Comments: 1754 Com21, Inc.
Category: Informational January 1995
IP over ATM Working Group's
Recommendations for the ATM Forum's Multiprotocol BOF
Version 1
Status of this Memo
This memo provides information for the Internet community. This memo
does not specify an Internet standard of any kind. Distribution of
this memo is unlimited.
1. Abstract
This document represents an initial list of requirements submitted to
the ATM Forum's Multiprotocol BOF for the operation of IP over ATM
networks as determined by the IETF IP over ATM Working Group and
other working groups. This RFC is issued for the benefit of community
members. The information contained in this document is accurate as
of the date of publication, but is subject to change. Subsequent
RFCs will reflect such changes.
The content of this memo was submitted by the IETF Liaison to the ATM
Forum as contribution number 94-0954 in the ATM Forum's documentation
process on 14 September 1994.
2. Notice
This contribution has been prepared to assist the ATM Forum. This
document is offered to the Forum as a basis for discussion between
the ATM Forum Multiprotocol BOF and the IETF. The statements are
subject to change in form and content after further study and
discussion. Specifically, the IETF reserves reserves the right to
add to, amend or modify the statements contained herein.
3. Introduction
The following is the charter statement from the Internet Engineering
Task Force's (IETF) IP over ATM Working Group (IPATM WG). It is
being reproduced here for the benefit of those in the ATM Forum who
may not be familiar with it:
"The IP over ATM Working Group will focus on the issues involved in
running internetworking protocols over Asynchronous Transfer Mode
(ATM) networks. The final goal for the Working Group is to produce
standards for the TCP/IP protocol suite and recommendations which
could be used by other internetworking protocol standards (e.g., ISO
CLNP and IEEE 802.2 Bridging).
The Working Group will initially develop experimental protocols for
encapsulation, multicasting, addressing, address resolution, call set
up, and network management to allow the operation of internetwork
protocols over an ATM network. The Working Group may later submit
these protocols for IETF standardization.
The Working Group will not develop physical layer standards for ATM.
These are well covered in other standards groups and do not need to
be addressed in this Group.
The Working Group will develop models of ATM internetworking
architectures. This will be used to guide the development of
specific IP over ATM protocols.
The Working Group will also develop and maintain a list of technical
unknowns that relate to internetworking over ATM. These will be used
to direct future work of the Working Group or be submitted to other
standards or research groups as appropriate.
The Working Group will coordinate its work with other relevant
standards bodies (e.g., ANSI T1S1.5) to insure that it does not
duplicate their work and that its work meshes well with other
activities in this area. The Working Group will select among ATM
protocol options (e.g., selection of an adaptation layer) and make
recommendations to the ATM standards bodies regarding the
requirements for internetworking over ATM where the current ATM
standards do not meet the needs of internetworking."
Historically, a large number of IETF IPATM WG participants are
employees of companies who are principal members of the ATM Forum.
Requirements between the two organizations have been communicated
informally by these participants. With the establishment of the ATM
Forum's Multiprotocol BOF activities, it has become prudent now to
document IETF requirements in a more formal fashion.
At the July 1994 meeting of the IETF in Toronto, Canada, a request
was presented to the IP over ATM Working Group by the ATM Forum
Liaison, Drew Perkins, for the working group to prepare a list of
requirements as input to the ATM Forum's Multiprotocol BOF
activities. This document is a response to that request.
4. List of Requirements for Consideration
4.1 Standardization & Logistics
- Formal communications between the IETF and the ATM Forum
should be made via IETF <> ATM Forum Liaison(s), specific
written communications (such as this document), and/or
presentations made at official IETF or ATM Forum meetings.
- IETF standards define how the TCP/IP protocol suite is defined,
deployed, and carried over specific network technologies,
including ATM networks [1][2][8].
- Any formal communications that affect the IETF standards
or processes must be made publicly available as the IETF is
a public international standards body. Ideally, such
communications should be written as Internet Drafts [1], the
IETF's equivalent to incoming contributions.
- We invite and encourage ATM Forum members to participate in
the IETF standards process. See [1], [2], and [8] for
information on how to participate.
4.2 IPv4 Encapsulation
- RFC 1483 [3] and RFC 1577 [4] define how IP is encapsulated
and carried over ATM networks. The IPATM WG requests that any
ATM Forum Multiprotocol work support these standards as
specified, and that any future changes to them be made via the
IETF standards process.
4.3 Routing
- RFC 1577 defines the default Logical IP Subnet (LIS) model.
- The IETF Routing over Large Clouds Working Group is developing
the Next Hop Resolution Protocol, which allows the incremental
optimization of routing (and subnets) by routing datagrams
over preferential ATM paths [9].
- The IETF IP over ATM Working Group will be working on the
next generation IP over ATM standards after RFC 1577 moves
from draft to proposed status. Requirements to the ATM
Forum will be forthcoming.
- ATM signaling should give an indication of connection
over LAN or WAN and include feedback of time vs byte
charging.
4.4 Security
- ATM signaling should support a user information element
that is used to convey security and authentication information
between IP members and applications. The IETF IPATM WG would
like to define the IP specific content of this IE.
4.5 Broadcast and Multicast
- The IPATM WG is currently discussing models of how best to map
IP multicast facilities onto ATM facilities. While this work is
preliminary, the IETF does support the ATM Forum's currently
planned multicasting enhancements, such as leaf-initiated joins
and support of multiple multicast congestion management
policies. A further list of requirements will be presented at a
later time.
4.6 Signaling and Addressing
- The IPATM WG is currently producing a specification for using
UNI 3.0 and 3.1 signaling to support RFCs 1483 and 1577. This
specification will be published as an informational reference
for UNI 3.0 signaling, and as a proposed standard for UNI 3.1
signaling following UNI 3.1's ratification and official
publication.
- IPv6 packets will include a Flow ID field intended to support
service classes in some way. Until the semantics of this field
are fully defined it is hard to say much, but presumably a soft
mapping between this and the VC to be used is desirable. A
further list of requirements will be presented at a later time.
- IPv6 addresses will be 16 bytes and there will likely be a
defined embedding of them inside 20-byte NSAP format. There will
also likely be a mapping of US-GOSIP-like NSAPs into IPv6
addresses (deleting the unuseful bytes), but that is still
controversial in the IPv6 discussions. A further list of
requirements will be presented at a later time.
4.7 Quality of Service, Performance, and Traffic Management
- ATM should support extremely bursty applications with
significant elasticity in their bandwidth demands.
- ATM should support elastic applications as defined in
RFC-1633 [7] very efficiently. That means enable high
bottleneck utilization while keeping delay reasonably bounded
(i.e., doubling delay wouldn't be terrible for elastic apps).
This should not be at the expense of delay sensitive classes
of service.
- ATM should provide a a class of service which strives to
cooperate with existing TCP congestion avoidance, thereby
explicitly providing support not only for directly ATM-attached
and -aware endstations, but also for endstations on LANs (or
using LAN Emulation) that are using current TCP implementations
and interconnected via ATM-attached bridges and routers.
- Predictive QoS should be supported in addition to guaranteed QoS
to support applications which are somewhat tolerant of delay
variation and low levels of loss.
- IP uses both point-to-point and point-to-multipoint (future)
connections. To satisfy IP's needs an ABR-like service
would need to be applicable to both types of connections [6].
- No specification of minimum or maximum bandwidths by the ATM
end-systems [6].
- As simple as possible [6].
- Full line-rate transmission over otherwise-idle links [6].
- When end-to-end delay through the network is less than 1 second,
the cell loss for AAL5 frames over an ABR-like service should be
on the order of 3 in 10**8 cells for 1500 byte frames, or 3 in
10**9 cells for 18 Kbyte frames [6].
5. Security Considerations
Security issues raised in this memo will be addressed by the IETF IP
over ATM Working Group and presented in subsequent updates to this
memo.
6. Acknowledgement
The basis of this memo is a summary of comments made on the email
discussion list of the IP over ATM Working Group. The contribution
was reviewed by Drew Perkins and Andy Malis as a sanity check before
submission to the ATM Forum.
7. References
[1] IETF Secretariat and G. Malkin, "The Tao of the IETF - A Guide
for New Attendees of the Internet Engineering Task Force",
FYI 17, RFC 1718, CNRI, Xylogics, Inc., November 1994.
[2] Internet Architecture Board, and Internet Engineering Steering
Group, "The Internet Standards Process -- Revision 2", RFC 1602,
IAB, IESG, March 1994.
[3] Heinanen, J., "Multiprotocol Encapsulation over ATM Adaptation
Layer 5", RFC 1483, Telecom Finland, July 1993.
[4] Laubach, M., "Classical IP and ARP over ATM", RFC 1577,
Hewlett-Packard Laboratories, January 1994.
[5] Deering, S., "Host Extensions for IP Multicasting", STD 5,
RFC 1112, Stanford University, August 1989.
[6] McCloghrie, K., "Lan-Emulation's Needs for Traffic Management",
ATM-Forum/94-0533, ATM Forum, June 1994.
[7] Braden, R., Clark, D., and S. Shenker, "Integrated Services
in the Internet Architecture: an Overview", RFC 1633,
USC/Information Sciences Institute, MIT, Xerox PARC, June 1994.
[8] Postel, J., Editor, "Internet Official Protocol Standards",
STD 1, RFC 1720, USC/Information Sciences Institute, July 1994.
[9] Malis, A., "Routing Over Large Clouds Liaison to the ATM Forum
Multiprotocol BOF", ATM-Forum/94-0766, ATM Forum,
September 1994.
8. IETF <> ATM Forum Liaison
Drew Perkins
FORE Systems, Inc.
174 Thornhill Road
Warrendale, PA 15086
Phone: (412) 772-6527
Fax: (412) 772-6500
Email: ddp@fore.com
9. Author's Address
Mark Laubach
Com21, Inc.
2113 Landings Drive
Mountain View, CA 94043
Phone: (415) 254-5882
Fax: (415) 254-5883
EMail: laubach@com21.com
AI ELI5
On this page
§1Background and Purpose
In the mid-1990s, Asynchronous Transfer Mode (ATM) was seen as a promising technology for high-speed networking, and both the IETF and the ATM Forum (an industry consortium) were working on how to run IP over ATM networks. Many IETF participants worked at companies that were also principal members of the ATM Forum, but communication between the two organizations had been informal.
In July 1994, the ATM Forum Liaison formally requested that the IETF's IP over ATM Working Group prepare a list of requirements as input to the ATM Forum's Multiprotocol BOF (Birds of a Feather) activities. This RFC is the response to that request, documenting the IETF's initial requirements for supporting IP over ATM networks.
The document emphasizes that IETF standards define how the TCP/IP protocol suite is deployed over specific network technologies, and that any changes affecting those standards must be made through the public IETF standards process.
Two big groups—one for the Internet and one for ATM phone cables—needed to talk about how to make them work together. This memo is what the Internet group asked for.
§2Core Encapsulation and Routing Requirements
The IETF required that any ATM Forum work support RFC 1483 (Multiprotocol Encapsulation over ATM Adaptation Layer 5) and RFC 1577 (Classical IP and ARP over ATM) as specified. These RFCs defined how IP packets are encapsulated and carried over ATM networks. Any future changes to these standards would have to go through the IETF standards process, not be decided by ATM Forum alone.
For routing, the RFC 1577 Logical IP Subnet (LIS) model defines the default behavior. The IETF was also developing the Next Hop Resolution Protocol (NHRP) in the Routing over Large Clouds Working Group, which would allow incremental optimization of routing by using preferred ATM paths.
The IETF made one operational request: ATM signaling should indicate whether a connection is over a LAN or WAN and provide feedback on time-based versus byte-based charging, so that end systems could make informed decisions about routing.
Use these specific ways to wrap up Internet packets in ATM format, don't change them without asking the Internet group, and tell us if you're charging by time or by how much data goes through.
§3Security, Multicast, and Addressing
For security, the IETF requested that ATM signaling support a user information element to convey security and authentication information between IP members and applications, with the IETF defining the IP-specific content of this element.
For broadcast and multicast, the IETF was still discussing how to map IP multicast facilities onto ATM facilities. However, the IETF supported the ATM Forum's planned multicast enhancements such as leaf-initiated joins and multiple multicast congestion management policies.
For addressing and signaling, the IETF was producing a specification for UNI 3.0 and 3.1 signaling to support the core encapsulation RFCs. Regarding IPv6 (then called IPng), the IETF noted that IPv6 packets would include a Flow ID field intended to support service classes, and that there would likely be a mapping between IPv6 addresses and NSAP (OSI) addresses within the 20-byte NSAP format.
Let IP add its own security tags, figure out how to do group messaging, and help IPv6 connect to ATM's address system.
§4Quality of Service and Performance
The IETF laid out expectations for how ATM should handle Internet traffic. ATM should efficiently support bursty, elastic applications (those whose bandwidth varies over time). For elastic applications as defined in RFC 1633, ATM should achieve high bottleneck utilization while keeping delay reasonably bounded, without compromising delay-sensitive service classes.
ATM should provide a service class that cooperates with the TCP congestion avoidance algorithm, supporting not only directly ATM-attached systems but also traditional LAN-attached systems using current TCP implementations interconnected via ATM-attached bridges and routers.
The IETF requested both predictive QoS (for applications tolerant of delay variation and low loss) and guaranteed QoS. Since IP uses both point-to-point and multipoint connections, an ABR-like (Available Bit Rate) service would need to apply to both types. The IETF emphasized that specifications should be as simple as possible, provide full line-rate transmission over idle links, and achieve cell loss on the order of 3 in 10^8 cells for 1500-byte frames under typical conditions.
Handle both steady and bursty traffic fairly, work well with how the Internet learns to send data, and keep delays predictable for real-time uses.