Re: [Netslices] COMS BOF presentations

[Vishnu Ram <[e-mail]>]

Hi Jari,

Sorry for being late to comment.
i was at your "setting the scenes" presentation and I felt it was quite
well thought-out.
Thank you for that well-balanced view.

I do have the following comments to the presentation
slides-101-rtgwg-sessa-considerations-on-network-virtualization-and-slicing-00

1) slide-3: "Virtualization does not generally affect TCP/IP or
applications":
there are a whole bunch of things which you mention in the previously
"Virtual networks, network function virtualization, software-defined
networking, service chaining" etc.
These definitely seems to affect the applications at least. Especially,
they seems to affect certain deployment assumptions
and operational parameters of applications.
In fact, to keep these apps "un-affected", developers (like me) have to do
extra work :-) on top of such virtualized infra that you mention.

2) slide-4: "Virtualization Tech @ IETF".
Having observed both IETF and non-IETF work for some while,
IMO, it would be good to look at an overall picture of where/how these are
consumed/produced.
Eg: what are the opensource communities doing with these works from IETF?
How are they using them?
Do we tie well with what say ETSI is doing in this space?
You do mention this in slide 10: "How our different pieces fit together": I
would suggest to broaden the definition of "our".

3) slide-5: "Architectural Observations 1 : In many cases, this replaces
the need for protocol mechanisms."
I wonder if I agree.
How about this view: due to the increasing role of software, services need
interoperability and hence the interface between them
need standards.
It would be interesting to hear your views.

4) slide-8: "Architectural Observations 3": "Merely the models + magic
software? ".
Now, this might not be an architectural issue,
but given how important these models are, dont you think it matters the
process and place where such models themselves are developed?
In that sense, I agree with you when you say in slide 10 that Data model
development should be a continued topic @ IETF.
But again, i think data models by themselves are not sufficient, and will
not stand alone, the corresponding interfaces and management of these
have to be worked upon.

5) slide-9:"IETF probably wants to do tools that work across industry as
opposed to only for 5G"
What would you say is an example, other than 5G, which needs this kind of
work?


thanks,
Vishnu.



2018-03-21 20:55 GMT+05:30 Jari Arkko <[e-mail]>:

> I’m on the agenda under “setting the scene”. I’m working
> on my presentation still, but if people are interested in what
> direction I plan to take it, I held a similarly themed earlier in
> the week in RTGWG, the slides for that were at
>
>     https://datatracker.ietf.org/meeting/101/materials/slides-
> 101-rtgwg-sessa-considerations-on-network-virtualization-and-slicing-00
>
> Tomorrow I plan to take this a bit more focused direction
> related to 5G slicing. Comments and suggestions much
> appreciated.
>
> Jari
>
> _______________________________________________
> Netslices mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/netslices
>



-- 

thx
Vishnu