NMRG Session #1 09:04 : Meeting start - 36 participants in the room, none present on jabber Introduction by Lisandro Granville - Note well  - Objective of the meeting:   - Learn from experience in autonomic management   - Build an NMRG/IRTF definition of terms - Agenda bashing Presentation #1: Sheng Jiang & Brian Carpenter - Network Configuration Negotiation Problem Statement and Requirements  The presentation promotes the need to manage any information available without human intervention. The major idea is to build a system that allows « plug and play for ISPs in the core network » The spreaders present a proposal and a generic protocol to support the idea of configuration by negotiation in multiple directions: between up and down devices, between peers, between networks. An example on IP address blocks negotiation among devices in a network illustrates the proposed system. Recommendations: 1/  Extend the investigation with a in depth study on the applicability of existing protocols in the IETF for the negotiation phase.  Presentation #2 - Jefferson Nobre - P2P for service level agreement violation The presentation proposes an autonomic scheme to dynamically provision active measurement probes in IP networks. Based on the fact that active measurement is expensive, the work investigates autonomic coordination of probe activation through different activation strategies: random, local, remote, combined, coordinated. No particular requirement is placed on existing nodes, especially it is not requested that every network node hosts a probe. Recommendations: 1/ Investigate the coordination function in case of topology change in the underlying network 2/ While probe selection is an interesting approach, a first step would be to decide where to put probes in the network and deploy them (optimal placement). Provide a comparison of both approaches 3/ The presented work assumes the metrics use basic tools. However, some tools today offer very advance adaptation facilities (e.g. RADAR). It is needed to take these facilities into account in the system (e.g. sampling capabilities) 4/ Clarify how these probes and their measurements can be validated by a lawyer, i.e. if the probes tells a constraint was never violated, then how can this be proven.  Presentation #3 - Michael Behringer - A Framework for autonomic networking A presentation of the content of draft (draft-behringer-autonomic-network-framework-01.txt) http://tools.ietf.org/html/draft-behringer-autonomic-network-framework-01, providing definitions and stating  basic principles of autonomic networking.  The document raised a very positive discussion in the session.  Recommendations :  1/ Make the definition part of the document an RG document.  2/ Beyond discovery and messaging protocols, identify functions that can be factorized in the autonomic plane 3/ Work on a draft that links existing IETF protocols to autonomic functions. Presentation #5 Laurent Ciavaglia - A running implementation of an autonomic network framework The presentation is focused on a framework around autonomic functions with a focus on the NEM and NEM skin concepts developed within the Universal Management Framework (UMF). A link to the framework described in the previous presentation is also included. The framework is being implemented and will be made available to the community to experiment its applicability to IETF protocols. Recommendation: 1/ Investigate the distribution of the UMF and its impact on performance 2/ Investigate a mapping with the framework defined in draft-behringer-autonomic-network-framework-01.txt NMRG - SESSION #2 (11h30-12h20) Presentation #1 Michael Behringer - Bootstrapping trust in a Homenet A method to bootstrap trust using a trust anchor in home networks using vendor certificates is presented. The proposed trust system is the first step to enable autonomic functions deployment. Recommendations: 1/ Clarify what functions are built on top of the protocols 2/ Investigate the possibility to have these protocols operate as generic containers to carry as many different scenarios as possible. NMRG Session eon Autonomics - Discussion Several  frameworks have been proposed in the session and some are published in other foras. A potential to integrate them at naming and definitions level is identified. A general agreement on the soundness of the work is reached but use cases now need to emerge, be presented and discussed to assess the value of a potential framework. The problem solved and the impossibility to solve the same problems with existing protocols need to be described as well. Existing approaches should be reconsidered in that context (e.g. the work on autonomic renumbering by Beck & all). Use cases will more likely be in the cross-domain area. Experiments need to be properly designed and documented. It is requested that any autonomic function helps system administrators to ease Initially presented as a framework, the work should rather move towards guidelines for autonomic networks and these guidelines should come out of the NMRG. A first step is to agree upon and move forward on the design principles  The RG proposal to be validated on the list are to work on 3 RG documents:    - One on definitions and concepts (including guidelines to reporting case studies)    - One on the Design goals and principles     - One on functions. A fourth document that, in coordination with the SDN WG, provides a common definition of terms regarding what is management and what is control, what gain an autonomic function brings but also what loss might be of interest as well. The NCRG might help in providing help determining in which direction we might be going Next NRMG Meeting : Proposed for London with a focus on the 3 documents.