IETF 46 ADSLMIB Minutes Mike Sneed chair, Dave Allan co-chair Agenda: Agenda bashing Administrivia and status ADSL Forum Update Draft-ietf-adslmib-adsl2-02.txt Draft-turner-adslmib-cpeautoconf-00.txt Path forward Started at 3:45 Added RFC 2662 implementation issues and profile semantics to discussion topics Administrivia Noted that we are slightly behind schedule on the supplemntary MIB ADSLF Update: Gigi Kharmous-Edwards Mode objects The four mode objects - line mode atuc capable ß current - line mode atuc enable ß need this one - line mode atur capable ß is this required? - line mode atuc current ß current Should be compatible with G.HS Atur capable. Agreed to take to the list "do chipsets permit the ATU-R capability to be obtained?" Power management: Use ifOperStatus for distinguishing between ON/OFF, we also need a status bit to determine when in the on Statte whether the ATU0C is in low or normal power state. Unavailable seconds is mutually exclusive of retrain and failed retrain seconds. G.HS Dual mode issue Dual mode will require a second profile pointer to be used only when dual mode is set. If single mode, then base profile covers G.DMT or G.LITE. If dual mode, then G.LITE moves to the overflow profile. Mike S. Will create a problem with static profile indexes! Need to add a hack on top of the I-F index hack. Only choices, do another table with additional index, or do separate table essentially identical to the first one. Agreed to take the problem off line. 'n' dimensions, dual mode, static mode. Draft-ietf-adslmib-adsl2-00.txt/Faye Ly Revised draft missed the deadline, Changes Resolved issues: - transmission capability C R - clarified unavailable seconds vs. known impairments - line mode capability and actual - lots of typos and MIB syntax errors Current status - were no issues outstanding until previous presentation - MIB covers G.997.1 Suggestion that we rename the MIB to G.lite , no obvious objections. Dave raised the possibility of handling, good idea to retrain vs. absolutely must retrain SNR margins for voice services. RFC 2662 implementation issues Synchronization of profile changes across systems. MIB does not state if profile can be reassigned while a port is active. Faye: Found a lot of parameters in the current table cannot be supported by the chipsets. Mike: don't want to do an object by object comparison as vendors may not have caught up to the MIB As for revving RFC2662, we need two independent interoperable implementations of all the objects. Clarifications will be deferred until we push 2662 to draft status. We can rev at proposed status anytime. Draft-turner-adslmib-cpeautoconf-00.txt Randy Turner – 2Wire Purpose is bounce ideas and get feedback. CPE is auto-configured from the network. Related to ADSLF work. First cut from l2/l3 perspective is getting connectivity. Last conf call cut was to provide network layer connectivity to one or more services. Approach is independent of the service, or the number of protocol layers between the subscriber and the service. Model should also support frame services. Once we have an IP pipe to the service, its up to service provisioning mechanisms to provision the service (be it audio-video etc.). If we stop at the network layer, there can be incremental service additions. "Service specific provisioning". Outlined some requirements, security. Suggests that the ADSLF model may be pull. (similar to cable). Capability discovery (how do you know you can support a service). This draft has not been presented to the ADSLF. Suggestion that the ADSLF complete it's work and bring it in as a BOF. This group is L1 issues. Are you deploying SNMPv3? Is it in production? We're not shipping yet. Path forward - wrap up extensions mib, main thing left is SNR and profile issues. - looking to be complete in December. - Randy's proposal will gain focus at the ADSLF. Bert suggests updating our milestones, adding Randy to charter is separate discussion item. - see Configuration Management BOF