Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: [EFM-OAM] Notes from yesterday's call




Matt, one comment below:

At 04:19 PM 4/25/2003 -0400, Matt Squire wrote:


>Here are my notes from yesterday's call.  Most of the time was spent 
>walking thru Clause 30. Touched on some 22/36/45 things but not too much.
>
>Attendees:  David Martin, David Law, Floyd Gerhardt, Barry O'Mahoney, Matt 
>Squire
>
>- Matt
>
>=============================
>
>Clause30 Title.  The current title doesn't address the new PHYs we now 
>have.  Should probably just call it "Management".
>
>Action.  Matt to make comment.
>
>=============================
>
>Figs 30-3/30-4.  Generally thought to be duplicates and that we can merge 
>them.
>
>Action.  DavidL to comment.
>
>
>=============================
>
>aSymbolErrorDuringCarrier.  It wasn't obvious to those on the call how to 
>interpret this attribute for copper (and maybe PON).
>
>Action.  Ask those that know more how copper/P2MP fit into this.
>
>=============================
>
>Profiles.  We asked if there were a way to have 10PASS-TS and 2PASS-TL use 
>the same profile management structure.  After some discussion, it didn't 
>look like we could simplify it any.
>
>=============================
>
>Events.  Lots of discussion on the events.  First, we decided we need to 
>have counters for the number of events, not just for the number of event 
>PDUs (each PDU can contain multiple events of different types).  Second, 
>are unconfortable with the current C30 handling for events, where the 
>latest received event info is an attribute.  Given that this information 
>can change multiple times per second, its quite possible that the changes 
>would be missed.  So it was suggested that instead of keeping the latest 
>PDU info, we should only keep counters.  Seemed like people wanted to 
>think a little bit about that one.

Did anyone bring up Jonathon Thatcher's parallel counter idea (keeps a 
running count, overflowing) from his comments on D1.3?  Discussion was on 
the reflector in late February 2003.  Seems that could be a way to maintain 
an accurate total count if we think we might miss an update, but maybe I'm 
misinterpreting the concern above?


>  Finally, it was pointed out that vendor specific event types are not 
> covered and probably should be.
>
>Action:  Matt to comment on event counters.  Everyone to think about 
>changing C30 event management philosophy to only counters rather than 
>"last seen" event.
>Someone <forget if we had a volunteer> to comment about attribute(s) 
>vendor extension events.

Thanks,

Brian