Hi Kireeti,

A couple of responses below with LK>.

 - Larry

From: Kireeti Kompella 
<[email protected]<mailto:[email protected]>>
Date: Friday, September 28, 2012 9:21 AM
To: Cisco Employee <[email protected]<mailto:[email protected]>>
Cc: Kireeti Kompella 
<[email protected]<mailto:[email protected]>>, 
"[email protected]<mailto:[email protected]>" <[email protected]<mailto:[email protected]>>
Subject: Re: [nvo3] What is CUG model was RE: Push or pull?

Hi Larry,

On Sep 28, 2012, at 08:39 , "Larry Kreeger (kreeger)" 
<[email protected]<mailto:[email protected]>> wrote:

I think Mehmet has a subtle, but valid point about the term CUG.  A CUG is a 
group of users, while a VN provides connectivity between those users.

More than that, a VN may also provide connectivity to other entities (storage, 
appliances).

LK> OK, so you are differentiating "users" of a CUG (presumably TES) from 
shared "services" provided by the cloud service provider.  Do you think we need 
some terminology for this differentiation or not?

So, good: the notions of CUG and VN are close in many aspects; however, VNs are 
what we want, and don't carry baggage.  Let's define VNs, then use them 
consistently across NVO3 documents.

LK> I agree!
There is one other aspect of a VNs that we had previously discussed, which is 
that TES connected to different VNs can use overlapping addresses.

Certainly, VNs belonging to different "customers" must support overlapping 
addresses.  VNs belonging to the same customer may or may not have overlapping 
addresses, depending on how the customer chooses to allocate addresses.

This is an aspect that goes beyond what users can communicate with each other 
to what addresses they can use.  Presumably, if one were to allow two members 
of different CUGs to communicate (by policy), then the addresses used by the 
members of the two CUGs better not overlap.  However, if a gateway is used to 
interconnect two VNs, then that gateway could also provide a NAT function to 
allow communication between TES on the two VNs even if they overlap.  So, 
perhaps address overlap is another distinction between VNs and CUGs?

This leads me to a question about using "policy" (without a gateway) to allow 
two CUGs to communicate.  Can this "policy" also define NAT to allow 
overlapping addresses between two CUGs?

Policy, whether via a gateway or otherwise, includes NAT.  Whether one has 
gateways or not is up to implementation and deployment.

LK> OK, just saying policy is general enough to include NAT…so let me ask my 
real question.  Does the notion of Policy as used by many in this group who 
work on BGP based Lx-VPN (using RT import/export), include the notion of NAT?

Thanks, Larry

 I now pass the mike to Aldrin :-)

Kireeti.

Thanks, Larry

From: <Toy>, Mehmet 
<[email protected]<mailto:[email protected]>>
Date: Friday, September 28, 2012 7:17 AM
To: Kireeti Kompella 
<[email protected]<mailto:[email protected]>>, Thomas Narten 
<[email protected]<mailto:[email protected]>>
Cc: "[email protected]<mailto:[email protected]>" <[email protected]<mailto:[email protected]>>
Subject: Re: [nvo3] What is CUG model was RE: Push or pull?

CUG term has been used since the days of X.25.
To me CUG only represents  entities/users that are on the network (i.e. VN), 
not the network or a set of connections making the virtual network.
On the other hand, VN represents the network or a set of connections making the 
virtual network, but not the users on the network.

Mehmet

From:[email protected]<mailto:[email protected]> 
[mailto:[email protected]] On Behalf Of Kireeti Kompella
Sent: Friday, September 28, 2012 9:36 AM
To: Thomas Narten
Cc: [email protected]<mailto:[email protected]>
Subject: Re: [nvo3] What is CUG model was RE: Push or pull?

Hi Thomas,

On Sep 28, 2012, at 6:09, Thomas Narten 
<[email protected]<mailto:[email protected]>> wrote:
This discussion is interesting, but I still have not seen an answer to
the question: what is (or is there) a difference between a VN and an
CUG?

No difference.


If there is no difference, that would be great, because then we can
use the terms interchangably. If there are differences, we need to
understand what those differences are, or there will be confusion in
our discussions.

For clarity, we should define VN in some draft (framework?), state that VN and 
CUG are the same, but stick to using one term, preferably VN.

Note that Wikipedia defines CUG as follows:


Closed User Groups are groups of GSM<http://en.wikipedia.org/wiki/GSM> mobile 
telephone<http://en.wikipedia.org/wiki/Mobile_telephone> subscribers who can 
only make calls and receive calls from members within the group. Any other 
calls would be rejected.


The definition is pretty close, but it goes to show that CUG comes with 
overtones that may confuse people.

One refinement might be to define L2 and L3 VNs off the base definition of VN.


Yakov Rekhter <[email protected]<mailto:[email protected]>> writes:


There is a definition of L2-based CUG in
draft-rekhter-nvo3-vm-mobility-issues

Thomas, the reference I gave below about where L2-CUG is defined is wrong -- 
not the EVPN draft, but the one above.

<rest snipped>

Kireeti

_______________________________________________
nvo3 mailing list
[email protected]<mailto:[email protected]>
https://www.ietf.org/mailman/listinfo/nvo3

_______________________________________________
nvo3 mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/nvo3

Reply via email to