Network Management Area Director(s): o James Davin: jrd@ptt.lcs.mit.edu Area Summary reported by James Davin/MIT At the Santa Fe meeting of the IETF, six working groups of the Network Management Area held one or more sessions throughout the week. Also, two Birds of a Feather sessions were held. The SNMP Network Management Directorate reviewed six MIB specifications that had been recently reported out of working groups. Three of these were products of the Character MIB Working Group: the MIB for character stream devices, the MIB for parallel-printer-like hardware devices, the MIB for RS232-like devices. Also reviewed were the IP Forwarding MIB produced by the Router Requirements Working Group, the Frame Relay MIB produced by the IPLPDN Working Group, and the SMDS Interface Protocol MIB produced by the SNMP Working Group. The IESG announced its intention to consider these MIBs as candidates for Proposed Standard status after final text is available in the Internet-Drafts repository. In addition, the Directorate discussed the problem of representing elaborate protocol stacks using the abstractions provided by the ``interfaces'' Group of MIB 2. The Directorate discussion was premised on the notion that is implicit in MIB 2 that an ``interface'' object is only used to represent protocol entities below the internetwork (e.g., IP) layer. The problem addressed has arisen in any number of Working Group discussions: although the interfaces Group in MIB 2 is a convenient abstraction for managers, it doesn't support specific transmission media or elaborate protocol stacks that may involve both downward and upward multiplexing. The Directorate discussion came to three conclusions: 1. Every entry in a media-specific MIB table is paired one-to-one with a single entry in the interfaces table of MIB 2. The media-specific entry can be reached from the generic interfaces table entry by using information in the ifType object together with information in the ifIndex object. 2. Media-specific MIB table entries can (and often do) include ``pointer'' information that represents user-service relations among entities in a more or less elaborate protocol stack below the internetwork layer. This pointer information variously takes the form of OBJECT IDENTIFIER values (as in the Character MIB) or combinations of OBJECT IDENTIFIER and INTEGER values. 3. If every protocol entity below the internetwork layer is 1 represented by an entry in the MIB 2 interfaces table, then all possible user-service relations among such entities may be concisely represented as a set of ordered pairs of ifIndex values. A simple MIB to represent such a set of ordered pairs was deemed desirable. A document presenting these conclusions in greater detail will be prepared as a basis for broader discussion of this problem. X.25 Management Information Base (x25mib) The Working Group met to consider three documents: one that instruments X.25 link-layer functionality, one that instruments X.25 layer 3 functionality, and one that instruments convergence functions necessary to run IP over X.25. At this meeting, the Working Group decided that the scope of instrumentation in the link-layer MIB will be confined to the LAPB protocol. The Working Group also concluded that the objects in these MIBs should be reviewed for actual usefulness in managing networks and that some pruning or alteration in conformance posture may be desirable. The Working Group noted that the IPLPDN Working Group was contemplating a revision to RFC 877 and decided to monitor that activity to determine if it may warrant revision to the IP/X.25 convergence MIB. The Group also discussed at some length the problems of representing X.25 protocol stacks in MIBs and suggested that the SNMP Directorate might pay some attention to this problem. Remote LAN Monitoring (rmonmib) This Working Group met informally to discuss implementation experience with the recently published RMON MIB. At the suggestion of members who had attended the Birds of a Feather session on SNMP Device Discovery earlier in the week, the Working Group spent some time discussing ways in which RMON technology could be applied to the device discovery problem. The meeting also recommended that a new working group be formed to address extensions of the RMON MIB for Token Ring media. IEEE 802.3 Hub MIB (hubmib) This Working Group met to discuss the current draft of an SNMP MIB for 802.3 Repeater devices. The Chair reported on IEEE reaction to this first draft of the SNMP MIB. A presentation was made on ideas for a ``Chassis MIB'' that is useful in instrumenting communications products that encompass multiple devices. As a result of this presentation, the Working Group concluded that its repeater MIB need not accommodate multiple repeater devices as this need was better addressed by the notion of a Chassis MIB. The Working Group recommended that effort be applied to development of the Chassis MIB ideas. Internet Accounting (acct) This Working Group met in two sessions during the Santa Fe IETF meeting. 2 The first session reviewed the Internet Accounting Background document (RFC 1272). Some time was spent bringing newcomers up to date with the Working Group's purpose and efforts. New attendees brought fresh perspectives and offered many comments, criticisms, and suggestions that will be incorporated into either a new version of the RFC or into follow-on documents. The second session was spent in discussion of the latest draft of the Internet Accounting Architecture. Although this document has existed for several months now and has undergone three or four extensive revisions, it still needs work, both in form and content. The stated scope of the document was tightened. The Internet Accounting model and its difference from the OSI accounting model was more clearly defined. A decision was made to combine the metering services document (formerly to be separate) with the architecture document. A decision was made to announce the Working Group's intention to produce a draft MIB document before its work is concluded. Discussion of the architecture document will continue with a view to advancing it to the status of Internet Draft by the next IETF conference. Simple Network Management Protocol (snmp) This Working Group met briefly in Santa Fe to conclude its business. The only item of outstanding business was the resolution of issues surrounding the Ethernet MIB. The Working Group Chair reviewed the course of action that had been previously discussed on the mailing list. With the formation of the Ethernet MIB Working Group to resolve outstanding issues, the SNMP Working Group adjourned and disbanded. The scheduled time that remained after adjournment of the SNMP Working Group was devoted to the first meeting of the new Ethernet MIB Group. Ethernet MIB (ethermib) The Ethernet MIB Working Group met for the first time in Santa Fe to begin its resolution of outstanding issues in the Ethernet MIB. The Working Group Charter was presented and interpreted by the Chair. The Working Group decided to omit from the current version of the MIB the language that dissociates conformance to the standard from actual implementation of the relevant objects. The Working Group felt that resolution of the issues required a more straightforward strategy that ties implementation requirements to particular operating environments. The Working Group also decided that distinctions between 802.3 and Ethernet environments could be a useful principle in articulating conformance requirements. The Group also agreed that distinctions between hardware and software implementations of MAC layer functions would also be an important consideration. SNMP MIB Compiler (mibcomp) A Birds of a Feather session on SNMP MIB Compiler technology was conducted by Dave Perkins of Synoptics. Dave presented his recent work on MIB compiler technology and explained how it could be valuable both in syntax checking of MIB documents and as a tool to support development of SNMP agents. 3 SNMP Device Discovery (devdisc) A Birds of a Feather session on SNMP Device Discovery was conducted by Fred Baker of ACC. Much time was spent in this session attempting unsuccessfully to formulate an adequate definition of the problem. The session articulated some ideas on how remote monitoring technology could be applied to the device discovery problem, and these were subsequently presented to the RMON MIB Working Group for its consideration. 4