Summary

  • RFC 3468 documents a February 2003 MPLS Working Group consensus to focus traffic-engineering signalling work on RSVP-TE and undertake no new CR-LDP work. It says the recommendations were accepted by the IESG, but it is an Informational record of that bounded process choice—not an Internet Standard or a network command.
  • The same RFC preserves separate facts that are often collapsed: CR-LDP’s document status and referenceability, individual contributions, product support, operator configuration, installed LSP state and service observation. A charter direction is not a receipt for any of those later layers.

A work programme is not a network configuration

The title of RFC 3468 is unusually direct: it documents the MPLS Working Group decision on MPLS signalling protocols. The abstract says the group reached consensus to focus its efforts on RSVP-TE for traffic-engineering applications and to undertake no new efforts relating to CR-LDP. Section 1 says the consensus was accepted by the IESG. Those are meaningful institutional facts. They identify which work a particular IETF group would advance.

They do not identify what a carrier ran. A working group can choose where it spends review time, drafts, charter space and standards-track attention. An implementation still decides which functions it supports. An operator still chooses a protocol, configures it, provisions a path and decides whether its own conditions permit use. A forwarding plane still has to show installed state; an OAM system still has to show what it observed; a service measurement still has to show the result at the relevant boundary.

Reading the first decision as if it completed all later ones makes the historical record sound stronger than it is. “The MPLS WG focused on RSVP-TE” cannot honestly become “this network used RSVP-TE,” much less “this LSP carried traffic” or “this customer received the intended service.” Every arrow crosses a different owner, time and evidence surface.

Referenceability survives a change of focus

RFC 3468 is careful about what its consensus did not do. It says there were no plans to progress RFC 3212 beyond Proposed Standard, but says no other action was taken regarding the status of RFC 3212 or RFCs that specified extensions to it. In its discussion of other standards organisations, it says CR-LDP needed to remain referenceable on the standards track for the foreseeable future. Keeping it referenceable did not mean further progression or new MPLS WG work was required.

This is a small but disciplined distinction. A group can stop creating new work in one area without rewriting the status of the existing document, and an external organisation can retain a reference while changing it at its own pace. The RFC records those separations instead of pretending there is one universal switch labelled “chosen” or “obsolete.”

The second phase makes the boundary sharper. The proposed action would change the MPLS WG charter; it would not change CR-LDP document status or hinder individual contributions. The discussion says continuing work was discouraged, not prohibited. That is a decision about a collective agenda, not ownership of all possible technical work or an order to every user of a reference.

A dated survey is not a present census

RFC 3468 records a June 2002 implementation survey with 22 responses: 21 implementations included GMPLS signalling based on RFC 3209, while three included signalling based on RFC 3212. The document uses that historical evidence to explain why the discussion occurred. It does not offer a current market census, identify products, disclose configuration defaults or demonstrate traffic in a particular network.

That distinction prevents a common evidentiary shortcut. A survey response can establish that a respondent reported something within the study’s terms and date. It does not establish that a feature remained supported, enabled or selected years later. It also does not establish an installed path, a successful reservation, policy compliance, capacity, loss, latency or a customer outcome. A historical implementation ratio is not an operational dashboard.

Keep the receipts with their owners

For a present technical claim, the necessary receipts remain local. A product record can establish a version and stated support. Configuration and provisioning records can establish an operator’s intended selection. Control-plane logs can establish a signalling action. Forwarding state and counters can establish an installed and used path. OAM and service measurements can establish their own scoped observations. None is supplied by a 2003 record of a group’s priorities.

Heng Lu’s distinction between a minimum shared specification and localized future decisions is useful here without adding facts to the RFC. A shared coordination decision can reduce duplicated work and preserve a reference boundary. It cannot make the later local choices real merely by being published. The value of RFC 3468 is not that it commanded all future networks; it is that it stated exactly which institutional layer had decided what.

Attribution without borrowed authority

Andersson’s public IETF profile lists RFC 3468 and provides the public identity reference for the editorial portrait. The RFC names both Andersson and George Swallow. The defensible credit is bounded: they documented a collaborative consensus record.

The sources do not show that either person alone chose a network, controlled a carrier, owned the implementation survey, selected a later product roadmap or guaranteed a traffic-engineering outcome. Precise attribution is not a smaller claim; it is the condition that keeps the work of a standards group from being used as borrowed authority over people who must make and prove their own operational decisions.

Evidence limits

The sources establish a 2003 MPLS Working Group and IESG process outcome, the preserved status/referenceability boundary, the dated survey figures and joint attribution. They do not establish current RSVP-TE or CR-LDP adoption, vendor support, a named network configuration, an LSP, traffic, performance, outage, security property or service result.

Sources