[Ibnemo] IB-Nemo on MOnday

["Susan Hares" <[e-mail]>]

Hi all: 

 

We had a lively discussion from the participants below.   We will be meeting
at 8pm (20:00) Budapest meeting on mezzanine level

 

We received lots of good input including the following suggestion for
improvement to our slides: 

 

1)      Need to emphasize what is the value of Declarative.

 

The point that the "Connection" is the single command that the application
does to connect after context is loaded.   We should show the command to
turn the connection off.

 

It is a question whether we should show the expanded theory We could expand
the idea in theory to include storage with a command "store" or the
calculation - Compute. 

 

2)      Need to explain how the "context" is pass between application and
the network management slides.  

 

We pass the exact topology in node and host statements.  Could we load this
context from the network for hosts?  Do we allow all hosts within a network
to connect? 

 

Bert was surprised that if Network 1 could talk to Network 2, and Network 2
could talk to network 3 - that network 1 and network 3 could not talk. We
need to make it clear that the flows are specifically defined in the model.


 

Alexander Clemm asked why this was better than the learning from the
network.  This is a question we need to consider.  Can we learn the context
from the network?  How does the user signal to load the network?  How do we
make this easy?  

 

[Sue's comment]: we need to chat  via email or in person about whether we
can signal the network management system to learn the context from its local
area.   Rather than specify all of the hosts within a network subnet - can
we say "allow all".   

 

3)      Start Demo at the user level 

Need provide a simple node example from Pedro's slide, and then go to a demo
with the same names.  

 

4)       For the second demo, we can show how the infrastructure provider
can allow choices at the infrastructure level.   We can then expand one of
the network  boxes to include various specific hosts. 

 

5)      Scott Bradner was concerned that we not work against other
network-wide yang models.  We need to emphasize that we will utilize
network-wide yang models and provide feedback to those network-wide models.


 

6)      Scott Bradner commented that there may not be enough network-wide
yang models in IETF to test until 2016.   This comment underlines that the
November IETF is the right timing for the BOF.  By that time, the L3SM and
other network-wide yang models may be created by other WGs.  This point is
important to underscore for slides. 

 

I will send slides for the IB-Nemo meeting tonight.  I would appreciate any
feedback on the changes on Thursday so tune the slides further.   I will
send the slide to the list tonight.  Or if you'd like to join an in-person
review 

 

Pedro and Tianran and I will walk through the slides a half hour before the
meeting on Thursday.  

 

Attendees for the IB-Nemo meeting were: 

Bert Wijnens  - Huawei consultant 

Diego Lopez  - Telefonica 

Pedro Aranda - Telefonica 

Susan Hares - Huawei Consultant

Scott Bradner - Harvard U

Alexander Clemm - Cisco 

Michael Schwan - ALU

Deng Hui - China Mobile 

Peng Fan - China Mobile 

David Dolson - Sandvine 

Ignaus Bagdonas RS Ltd. 

One person from Cable Labs. 

 

13 people