[Nsi-wg] call for new functions in NSI

[Henrik Thostrup Jensen]

Hi

On Wed, 26 Jun 2013, Guy Roberts wrote:

> Now that NSI v2.0 is nearly out the door we should be looking ahead to the next phase of NSI.

Getting a protocol implemented, tested and integrated into a production 
infrastructure takes a significant time. I think we need be careful about 
how we approach "the next phase" as we don't want people halting NSI 
adoption and waiting for yet another release (ok NSI2 is significantly 
more useful and NSI1). Goals and timelines should perhaps be the focus of 
the next meeting (we tend to get lost on the hows).

> The NSI working group will  meet in Singapore in the week beginning 30th September to discuss the
> next steps for NSI – details of the meeting will be announced on this mailing list.

Does that mean we won't do anything at OGF39 in Madrid? I realize OGF39 in 
Madrid and Glif in Singapore are very close in time, so it might not make 
sense to have something at both.

> If you have any ideas for NSI enhancements please prepare a  presentation for that meeting.

I think we should also consider cleaning up some of the existing 
constructions in the current protocol. For instance the acks where 
conceived when some actions like provision and release could take a long 
time, due to activating the circuits, but they no longer have this role, 
and we could probably do without the acks. Simiarly I think we should also 
consider a protocol which is not based on request-reply, but truly 
message. Similarly I'd also like to kill off the correlationId and always 
have time be in UTC at the protocol. I'll try and come up with something 
for the meeting.


     Best regards, Henrik

  Henrik Thostrup Jensen <htj at nordu.net>
  Software Developer, NORDUnet