NML/NSI meet

--- Port Groups    TK questions bi-direction pg's vs vlan pg's

Does PortGroup represent multiple relations?

TK:  Is link aggregation a Port group?   JVDH: THis is an aggregation.

JS: dissertation about STPs...

FD:  look at parallel compound links

Discussion about parallel compound links as "lagged" or bonded links...

JS:  Propose we need more detailed discussion on port groups


--- Topology distribution mechanisms


central
git hub
http linking 
peer to peer

JS:  How does the current github scale?  can we include reachability?

JvdH:  no way to currently aggregate reachability information... could possibly establish aggregating NSAs
Needs more thought.


p2P is potentially promising for updating... but requires new protocol

--- Need user topology access...   JS: why?  WHy should user need a different means of learning topo than a network?

TK:  Should be an NSA at edge of NSI world...  "user" defers to first hop NSA.
JS:  Why not let "users" see what is advertised, and let middleware developers build apps that summarize in discipline specific manner...?

RHJ:  The providers will advert whatever they wish the world to see, and users will have to work with that.

JS: agreed.  And "user" will need to interpret as desired.

--- perfSONAR topo service

CG (Chin):  Can ps Topo service do for NSI?

JS: what is magnitude of diff re:  topo services itsef, vs the topology offered by the topo service?

CG: thinks the is an editor that checks conformance of topo... JvdH:  Not sure the NML insertion into existing toposervice is as simple...

JS:  propose action item to compare a) ps topo service vs NSI needs/requirements, b)  implications of differences of NMWG vs NML

Roman: current xpath vs xquery(?)  (questions to Chin about current ps Topo service/lookup service...)

summary of discussion:  There is need for detail analysis of current Lookup service, topo service, vs "new" lookup/topo service, vs an NSI TS requirement.    Also need to understand differnence in topology ontology generally.
---
TK:  Comment to JvdH:  NSI uses the term "service plane" rather than "control plane".  please fix comments.


--- RHJ
discussion with NM/NMC and documents...
review progress  with Martin and Jason.

Ask for input from docs from this group.

RHJ does not think an implementor could read current docs and make effective interface.