NMRG Session 13:04 : Meeting start - 55 participants in the room, 12 present on jabber Introduction by Lisandro Granville & Olivier Festor - Note well - Objective of the meeting: continuation of previous meetings on autonomic management - Presentation #1 - Michael Behringer - Autonomic Networking – Definitions and Design Goals (draft-irtf-nmrg-autonomic-network-definitions-01) A presentation of the content of draft, providing definitions and stating basic principles of autonomic networking. Michael highlights the differences between draft revisions 00 and 01. Michael asks for last call considering the document flow of IRTF. Benoit Claise asks about the logical operator between the self-* features for an autonomic solution: AND or OR. Michael states that it is currently a logical OR to be considered as autonomic. Benoit questions the importance of modularity in autonomics (is more like a deployment goal). Michael states that modularity is the outcome and a general principle which is not totally linked to Autonomics. Tom Eckart questions about the solution and deployment goals. Kostas Pentikousis questions the related work section, arguing that hundreds of papers have been published on autonomics and that this work should not be ignored. Kostas also says that GANA is not the first and unique architecture in autonomics. Oilivier Festor states that the proposal of the document is not to review the autonomic networking literature. Sunir Das asks if the solution is just for routers and also for clarification about the control and management plane. Bill questions about security trust among devices in an enterprise. He also comments that the topology is deployed after one starts talking. Kostas questions the depth in which one can go into standardization since usually the intelligence (which is the core here) is kept by vendor. Tom says that the draft should provide the minimum necessary to go to IETF. Brian Carpenter says the document is a reference of a autonomic network node. There is a discussion on whether to remove the reference model part or not. Recommendation: Review the draft considering the suggestions, then send LAST CALL. - Presentation #2 - Brian Carpenter - Gap Analysis for Autonomic Networking (draft-jiang-nmrg-an-gap-analysis-00) A presentation of the content of draft, considering a problem statement for an IP-based autonomic network that is mainly based on distributed network devices. Kostas Pentikousis argues about difference between autonomic and automatic considering the given examples. Michael states why the history and the gap analysis are important in the context of autonomic aspects of IP networks. Ridchardson asks for clarification about history. Kostas Pentikousis likes OSPF and IS-IS being considered autonomic and asks how the draft is related to I2RS. Michael comments about relation of gap analysis and SDN approaches, such as I2RS. Xing says most of the work is about IP-based gap analysis and whether can it be generalized. Benoit clarifies that the current focus is on IP. Xing questions about using IP as control plane. Brian talks about interaction about virtualization and autonomic and says there is no reason for not using layer 3 communication. Recommendation: LAST CALL - Presentation #3 - Jeferson Nobre - Autonomic Networking Use Case for Distributed Detection of SLA Violations (draft-irtf-nmrg-autonomic-sla-violation-detection-00) A presentation of the content of draft, an use case about the autonomic provisioning of active measurement probes in IP networks. Kostas Pentikousis questions if the use case is aimed at a single administrative domain since he thinks it will not work on multiple administrative domains. Jeferson says that the draft is agnostic about single/multiple domains, but in theory could be applied for multiple domains. Al Norton states that single measurements are not enough to violate an SLA and that the customer can "over blast" its interfaces. Thus, he suggests to look in more complex systems. Benoit says it looks like the use case is trying to find the best locations to put probes and the number of probes. Jeferson says that there is no question of deciding of where to put probes, operators do that. Benoit states that trying to determine where to put probes is not the right approach and is not how the ISPs work. Yaakov Stein suggests to look at the METRO Internet work. Besides, he says it is possible to find that one violates a SLA on a rather daily and monthly basis. Recommendations: no recommendation given at this stage Presentation #4 - Abdelkader Lahmadi - Information Elements for IPFIX Metering Process Location (draft-irtf-nmrg-location-ipfix-01) A presentation of the content of draft, which defines a set of Information Elements for IPFIX protocol for exporting location information of any device (both fixed and mobile) that acts as an IPFIX Flow Exporter. Olivier comments about the lack of interest in IPFIX for standardization and Benoit says it should be concern to IANA the assignment of information documents. Tom states that the IPFIX registration is just sufficient for geo needs. Brian talks about geo privacy relation of this work. Kostas questions why this work is not standardized in IPFIX. Benoit says that IPFIX is no longer accepts drafts and if the objective is to document information elements, a IANA number would be sufficient. Benoit also clarifies that it is necessary to define a complete template (full semantics) to put to IANA. Kostas says that formats should be taken from other places where they are defined in the IETF. Abdelkader says that this is the case. John says the proposal makes fully sense if it is focused on what and why it is necessary (e.g., which type of devices for which attributes, how the can be activated, etc). Recommendations: update the draft so as to focus on semantics and usage scenarios, while not focusing only on the information elements. %%% Lisandro emphasizes the invitation for participation on the UCAN BoF on Wednesday, where more use cases will be presented.