[RPRWG] PAH: Meeting minutes Aug 7/02

[George Suwala <gsuwala@xxxxxxxxx>]

Meeting minutes of Protection Ad Hoc (PAH) meeting:
Date/Time: August 07, 2002, 10:00 AM America/Los_Angeles

Those minutes go out to the whole reflector, subsequent PAH
working communication will only go to the CC: names plus
individuals who explicitly request to be copied (please send
the request to be copied to the CC: list only)

The meeting was attended by:
Daniel Zhu
Gal Mor 
Jason Fan
Kshitij Kumar 
Leon Bruckman 
Robert D. Love
Jim Kao
David James
John Lemon
George Suwala 

Next meeting: Mon Aug 19 at 9AM PST, Jim will send out
the announcement. Subsequent meetings will be on Mondays 9 AM PST.

Minutes taken by George Suwala and Jason Fan

PAH objective: Close issues and publish resolutions (revised
protection and topology clauses) by Sep 15. Also understand and document
partitioning  between Topology and Protection by Sep 1.

Short term objective: By Aug 15 email proposal from 2 subgroups:
1. Node addition/power up scenario - (Kshitij - subgroup leader)
2. Consistency checking (topology) - (Jim - subgroup leader) 

Issues resolved today:
- Multiple Signal Degrade: If there are more than one SD
in the ring, there are no wraps in the system (all SD stations unwrap
if they have wrapped already). Station detecting a SD wraps immediately
in the absence of any other SD.
- Ring wrapped status bit: We keep the wrapped status bit in
the header.The wrap status bit does not appear to be necessary for the protocol, 
but is useful for ring status/ring debugging, specially that nodes
may be reporting SD and be in either wrapped or not wrapped state.
- Group re-iterated that as stated in the draft, the topology and 
protection messages will not be wrapped

______________________________________________________________
Issues brought up today:

1. If ring is wrapped and a station which does not support wrap is added
(already contained in the original issues list). 
2. Node initialization: should a node send messages immediately on 
initialization and do neighbors react immediately? Kshitij already works on scenario/state machine and will send out proposal as per the short
term objective #1 above.
3. TTL - should be set to constant or set to ring size? If set to the ring
size problems may arise when a node is added/deleted and the
ring size is inconsistent among nodes.
This issue is owned by the data path group, unless we choose to 
have a separate TTL for protection.
4. Do we allow a station to enter a pass-through mode where it looks
like a cable and if so what impact does it have on topology/protection?
If there is no additional input on that topic the current Draft position
is not to cover it. In any case we may have optical switches which
will change the ring topology without introducing SF. As a result
the topo and protection protocols must be robust enough  to handle
such cases.
5. Should we produce C code reference? 
6. What are the configuration options for revertive/non revertive modes? 
Per span or station and/or class of service? Report non revertive recovered link 
as WTR or idle?  Report WTR for wrapping and idle for steering? Remove 
non revertive mode completely? Have user configure non-revertive through 
WTR set to infinity or through a separate bit?
7. As per OAH request, the issues list will also include consideration of how to 
handle capabilities reporting as related to class A reservations from each station 
on each span. Should this be done in a separate message?

______________________________________________________________
Agenda and issues list as per earlier email from Jim and Jason:

1. Verify that issues list is complete
2. Partition tasks as appropriate, 
3. Plan order of resolving the issues, 
4. Set up process ground rules

If you have any suggestion for agenda or contribution you want to make in the first meeting,
please let me know by 8/4 so that I can allocate time slot in the agenda.

As for the issues, Jason and Jim  have grouped them as the follows.

A. Determine partitioning of topology and protection functions between protocols 

A.1     Clearly identify and itemize interface between topology, protection, and other modules within the MAC
(may be impacted as issues in B and C are addressed as well) 

A.2     Are topology and protection separate from a protocol standpoint and/or from a messaging standpoint?
A.3     Clearly identify triggers for sending topology messages
A.4     Can protection distinguish, using only protection messaging, whether a newly connected station is a 
        new neighbor or the same as a previously connected neighbor

B. Address remaining topology-specific issues 

B.1 What actions, if any, are taken upon failure to receive topology messages? 

B.2 Criteria and algorithm for consistency checking of topology image

B.3 Get input from RAH on units and ranges of provisioned class A bandwidth information reported in topology messages

B.4 Generate revised topology clause for September/October mtg.

C. Address remaining protection-specific issues in the following suggested order:
Grp a:  
·       How uniform should the behavior of wrapping and steering be in terms of reporting of link status throughout the ring? 
·       Is the wrapping status bit required?
·       Insertion of steering station into an operating wrapping ring
·       Do multiple signal degrades result in a single wrap or no wraps?
Grp b:  
·       How do non-revertive modes work for steering and wrapping? 
·       Retransmission of messages: exponential backoff or fixed period, what are limits on period?
·       How is variability of fairness advertisement interval handled as related to the loss of keep-alive trigger?
Grp c:  
·       Generate revised protection clause for September/October mtg

___________________________________________________________________