CURRENT_MEETING_REPORT_ Reported by Bill Kwan/CrossComm Corporation Minutes of the SNA DLC Services MIB Working Group (SNADLC) The SNA DLC MIB Working Group met to review the status of the SDLC MIB and continue discussions on the LLC MIB. SDLC MIB The current draft of the SDLC MIB, ``Definitions of Managed Objects for SNA Data Link Control: SDLC'' (draft-ietf-snadlc-sdlc-mib-04.txt), was reviewed by the Network Management Area Directorate. The results were discussed. 1. It was recommended to replace the document references to reference numbers for easier future editing. Dave Perkins indicated that references within the SMI should still be document references because the information will be lost, without the surrounding text, when the MIB is pulled out. 2. Some minor editorial changes to relationship to ifGroup. 3. Use ifIndex instead of sdlcPortAdminIndex to index Port table. 4. Concerns regarding activity timeout being optional. It should be made mandatory or removed. Need additional comments from mailing list. 5. Should conformance statements allow for a MIN-ACCESS of read-only (it currently does not)? Need additional comments from mailing list. A new Internet-Draft will be resubmitted ASAP (requesting for Proposed Standard status), as soon as items 4 and 5 are resolved. LLC MIB The schedule is as follows: o Jul 94 - Review LLC Internet-Draft (30th IETF) o Aug 94 - Post LLC Internet-Draft revision o Aug 94 - E-mail review of LLC Internet-Draft o Sep 94 - Post LLC Internet-Draft revision o Oct 94 - Review LLC Internet-Draft (AIW6 - interim meeting) o Nov 94 - Post LLC Internet-Draft revision o Dec 94 - Achieve consensus on LLC Internet-Draft (31st IETF) o Dec 94 - Submit LLC Internet-Draft for NM Directorate review Open issues raised at the AIW5 (APPN Implementer's Workshop) interim meeting in June were discussed. o Indexing of Connections An issue was brought up that an ifIndex and a MAC address may not be one to one. Here are the different scenarios: - 1 local MAC per ifEntry - n local MAC per ifEntry (DLSw) - n ifEntry with same local MAC (Group Addr) It has been observed since the WG meeting that group addresses are only used connectionless traffic, so it should not be an issue. However, for the other case, it looks like we might need both ifIndex and localMAC for indexing. Ravi Subramanian(IBM) proposed using: ifIndex, localMAC, local SAP, remoteMAC, remoteSAP (in some order). Need more discussion on mailing list. o Dynamic nature of connections As dynamic connections may come up and down rapidly along with their corresponding oper entries, information may be lost and not kept around. Several alternatives were discussed: - No oper entries for dynamic connections. Instead, have some global (by Port, SAP?) objects which keep track of aggragate information on dynamic connections such as number of and summary statistics. This can also be the place to put the default parameters for dynamic connections. - Have an optional event/history table where inactive dynamic connection oper entries are moved to. Variation is to not have dynamic connections in the oper table, but just shows up in the event/history table when inactive. Need more discussion on mailing list. o Object naming convention - Decision to use llc (vs. dot2) as the prefix was reiterated. - All of the NCP parameter names need to be changed to match IEEE 802.2 terminology with references made to NCP names within the description. - States - use IEEE 802.2 states. o XID As decided at AIW5, IEEE and SNA XID will be kept separate. Still waiting for Alan Bartky's (Sync Research) and Peter Gayek's (IBM) submittals on XID parameters (what LLC parameters (if any) are negotiated in SNA XIDs). o Global parameters We decided not to have any global parameters in the last meeting. Dave Perkins mentioned that other MIBs (e.g., X.25) does have the concept of global parameters, in case we decide to change our minds in the future. o Traps - Discussed techniques of RFC1224 to minimize traps. * Feedback mechanism - prevent over reporting * Polled alerts - prevent under reporting +If doing this, Dave suggested to define a structured entry rather than use an OPAQUE alert type since it is not widely supported. +For connections, could use same event/history table as discussed relative to dynamic connections. - Should we use traps or history log for connection events? - Should we use traps or history log for port events? - Should we be able to control event reporting at port level (SAP or Global) ? - Further discussion is necessary. o Number of counters Due to the number of objects, we decided we needed to look at having a base and extended set of counters. Ravi (IBM) will post submission to mailing list. o Dependencies on sysUpTime Discussion on Dave Chen's (IBM) concern about having timestamps be sysUpTime. His concern is that the timestamps will be meaningless if the system recycle. His proposal is to have the timestamps be the time since the event occurred. As agreed at AIW5, Dave was supposed to submit a formal contribution. Although Dave was not at this meeting, some discussions still took place. The general consensus was that current usage is consistent with other MIBs and Dave's suggestion may make implementation more difficult due to synchronization issues. More discussion can occur after Dave submit his contribution. The other scheduled meetings were cancelled because they were not necessary. Action Items (From AIW5) o TimeStamp proposal - Dave Chen o XID (IEEE versus SNA) - Peter Gayek and Alan Bartky o Indexing (Group of SAPs) - Dave Chen New Action Item o Proposal on base and extended counters - Ravi Subramanian Next Steps Update the draft LLC MIB and continue discussions on the mailing list. The next interim meeting is in October at AIW6 (Raleigh).