Access Node Control Protocol (ANCP)

1
Access Node Control Protocol (ANCP)

• IETF 66, Montreal
• Wojciech Dec (wdec_at_cisco.com)
• Matthew Bocci (matthew.bocci_at_alcatel.co.uk)

2
Administrivia

• Blue Sheets
• Note takers Jabber Scribe
• Mailing List
• General Discussion ancp_at_ietf.org
• To Subscribe ancp-request_at_ietf.org
• In Body subscribe your_email_address
• Archive http//www.ietf.org/mail-archive/web/ancp
  /index.html

3
Agenda

• Introduction/agenda tweaking (Chairs)
• Working Group Charter (Chairs - 10 mins)
• ANCP Requirements (Stefaan de Cnodder - 15mins)
• GSMP extensions for layer2 control (L2C) Topology
  Discovery and Line Configuration (Sanjay Wadhwa
  - 15 mins)
• ANCP Graceful Restart Mechanism (Sanjay Wadhwa -
  10 mins)
• ANCP MIBs (Stefaan de Cnodder - 10 mins)
• Outstanding Work Items (Chairs - 30mins)

4
Charter Changes since the BoFhttp//www.ietf.org/
html.charters/ancp-charter.html

• Clarified terminology and WG name
• Clarified non-goals
• Set-up of VCs or generic Access Node Management.
• Firmed up security objectives
• Reliability Scalability
• Graceful restart
• Distribution of functionality

5
Revised Milestones

• May 2006 ANCP MIB Last Call
• Nov 2006 Accept WG I-D for ANCP Framework
  and Requirements
• Jan 2007 Accept WG I-D for Access Node
  Control Protocol (ANCP)
• Jan 2007 Framework and Requirements last call
  
• Mar 2007 Accept WG I-D for ANCP MIB
• Apr 2007 Access Node Control Protocol (ANCP)
  Last Call
• Jul 2007 Re-charter or conclude Working Group
  

6
Mailing list changes

• All subscribers to L2CP mailing list should have
  been moved to ancp_at_ietf.org
• l2cp_at_ietf.org has been closed
• Please use ancp_at_ietf.org for discussion relevant
  to the WG
• To subscribe
• Send mail with subject subscribe to
  ancp-request_at_ietf.org

7
ANCP Requirements / Framework

• Stefaan

8
GSMP Extensions

• Sanjay

9
ANCP Graceful Restart

• Sanjay

10
ANCP MIBs

• Stefaan

11
Outstanding Work Items

• Functional Partitioning and multiple controllers
• Light-weight transport protocol
• ANCP Protocol Security
• Multicast control

12
Functional Partitioning and multiple controllers

• What are the partitioning capabilities required?
• Q1 Controller redundancy? How many redundant
  controllers?
• Q2 Controller functional split?
• Q3 Do we envisage both to be required? Eg
  Redundant controller for QoS and single
  controller for OAM
• Q4 Is it envisaged that multiple controllers,
  possibly managed by different operators, would be
  controlling the same function?
• Q5 How is the ownership of a controlled port
  to be negotiated?
• Q6 What are the security implications of
  functional partitioning across organisations?

13
Light-weight transport protocol

• High level goal is to scale the number of ANCP
  controlled nodes without compromising the
  transactional capabilities
• Q1 What are the real transport requirements? Do
  all ANCP messages and interactions require the
  same mode of transport and message delivery, eg
  unicast reliable message delivery?
• Q2 Does the light weight protocol become the
  long term goal over TCP?
• Q3Do we look for an existing light-weight
  transport IETF protocol?
• Q4 Can SCTP be a valid alternative?

14
ANCP Protocol Security

• Minimally addressed in the current draft via
  pre-configured peer IP address based security at
  the moment.
• Transport protocol security is an option, but
  this might not be applicable for when an
  alternative transport is defined or might not
  cover ANCP risks
• ANCP Protocol security appears to require WG work
• Security requires from a threat model of ANCP and
  ANCP transport to drive the security
  requirements.
• Operational considerations also drive some
  security requirements ANCP aims to simply
  operations across organisational boundaries. Use
  of pre-shared keys could impact such
  simplification.

15
Multicast control

• Multicast control appears to encompass two main
  functions
• Controlling directly the per port multicast group
  replication or multicast data plane filtering on
  an AN
• Reporting to the NAS per port membership and
  stats
• Use-cases remain to be firmed up
• Need to clarify relationship with exiting
  multicast techniques and protocols
• Input from multicast WGs would be beneficial
• No intention to reinvent the wheel.