Thank you Sharon and other authors

I read this with great interest.  I have some questions though to better 
understand how this would work:

1) in the figure on page 6 what is the "Distribution-Center B"? Do you mean 
Data-Center B?

2) why do you need to co-locate Control-Agents with the SDN-xTRs?

3) Section 5 page 7: what is meant by LISP overlay in this context? Or do you 
mean SDN-outerlay instead here?

4) Section 5 page 7: for the selection of the best suited VNF you seem to use 
quite dynamic attributes, e.g. "health-load attributes". This seems to indicate 
a push type of mapping system. The current LISP mapping system is pull based. 
And if I am not mistaken that was the design choice.

5) Section 5.1 maybe you clarify in the terminology section what is 
"Function-Instance-EID" and "Function_EID"? Are they same of different?

6) The text suggest that Mapping system (also) subscribes the SDN-xTR to the 
Function-Instance-EID and Function EID to receive updates. How are these 
updates, I believe "health-load" attributes, distributed to Control Agents of 
SDN-xTRs? Maybe a push based mapping system is required? How much overhead 
traffic that would generate?

7) Can you, please clarify in the terminology what is App-Instance-Record and 
how it differs from Function ID and Function-Instance-ID? 


Best regards
Hannu  
 



-----Original Message-----
From: [email protected] [mailto:[email protected]] On Behalf Of ext 
Sharon Barkai
Sent: Thursday, May 23, 2013 5:59 AM
To: [email protected]
Subject: [lisp] updated draft-lisp for sdn-nfv

There is an updated draft-lisp for sdn scaling network function virtualization.
http://www.ietf.org/id/draft-barkai-lisp-nfv-01.txt

The draft shows how LISP enhances SDN from (1) concentrated,  (2) hop-by-hop, 
(3) push control model, to (1) distributed, (2) overlay, (3) pull-pub-sub 
model, able to perform dynamic mapping the right flows, in the right sequence, 
from mobile users, to elastic functions.

Main additions to draft-01 by Vina clarify the terminology around LISP-SDN and 
the LISP messages, as well as explain the SDN-XTR flow-rule (exact and best 
match) affinity determination and forwarding model.

It will be good if during the next wg meeting Dino and Fabio will further 
elaborate on the LISP-SDN use-cases and the SDN open-source platforms LISP is 
being inserted to. Additional explanations on why and how to enhance SDN with 
LISP can be found on the (2page) paper bellow: 


_______________________________________________
lisp mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/lisp

Reply via email to