What about the following naming proposal:

-          Virtual Network Endpoint Reachability Information Distribution 
Authority (VNERIDA)

It is a functional description  - as least AFAIK.
It leaves it open if the information distribution model is push or pull or 
mixed push&pull, uses BGP or DNS or LISP or you name it.
The pronounciation of the acronym is internationally fairly unambiguous, and it 
can be fairly easily pronounced.

It does not use the term “network virtualization mapping” which is IMHO too 
vague, as neither networks nor virtualizations get mapped, rather the overlay 
address of a network endpoint in a virtual network gets mapped to a network 
endpoint identity belonging to a network endpoint in an underlay network.  
Please note that there may be a requirement for a second mapping if said 
underlay network endpoint identity is not a locator type routable address.  In 
this case, this unterlay network endpoint identity must be mapped to a routable 
underlay network endpoint identity – presumably an IP address.

It emphasizes that this is all about network endpoints which are vertices in a 
network according to mathematical graph theory. Most other names proposed so 
far have been given the – mathematically speaking – wrong  impression, that 
this “so far nameless concept” may have anything to do with the arcs in a graph 
theory network.

But maybe I am wrong. May I ask that those please speak up who believe that  
there is indeed a requirement that this  “so far nameless concept” should  
indeed also be responsible for the distribution of  information on arcs, i.e. 
on underlay network topology changes which do not affect the location of an 
overlay network endpoint.


And while I am at it please note my opinion on the proposal to rename the term 
“VNIC”

This is a No-Go. I would strongly recommend to leave the term VNIC as is. The 
compute centric colleagues from “the other side” of virtualization have defined 
it to be able to abstract it as the virtual equivalent of a physical NIC in a 
virtual mashine, allowing all things possible with physical NICs in a server 
(including NIC-Bonding = LAG) to be done with virtual NICs of a VM.  We simply 
do not have the authority to change the meaning of that, because it may break 
virtualization concepts in the datacenter.

But we should define the tem VNIC from the network centric side.

My proposal:
A VNIC is a virtual network endpoint. One physical NIC may have multiple VNICs 
associated to it, each with its own MAC-Layer.

This definition also closes the loop to the network endpoint centric 
terminology proposal I made in the first part of this mail.

Lothar


Von: [email protected] [mailto:[email protected]] Im Auftrag von Jon 
Hudson
Gesendet: Mittwoch, 10. April 2013 10:10
An: Sharon; [email protected]
Cc: Truman Boyes; Lou Berger; Larry Kreeger (kreeger)
Betreff: Re: [nvo3] NVO3 Terminology changes


My opinion: Either the words the acronym is representing need to be descriptive 
& clear. Which I think this one does well.

NVMA - Network Virtualization Mapping Authority

Or it needs to be a nice acronym that rolls easily of the tongue and so will be 
freely accepted into techlish. I think this one could fill that nicely.

NVO - Network Virtualization Orchestrator/Oracle

For a combination of both goals maybe something like this

OMA - Overlay Mapping Authority

Or keep it simple with a beloved tech culture reference ;-)

MCP - Mapping Control Program

End of line.


On Apr 9, 2013, at 9:59 PM, Sharon 
<[email protected]<mailto:[email protected]>> wrote:
NVMA +1

Sent from my iPhone 650 492 0794

On Apr 9, 2013, at 9:47 PM, Truman Boyes 
<[email protected]<mailto:[email protected]>> wrote:
Hi,

I think the below two suggestions provide the clearest terminology and 
description for the function:

NVA - Network Virtualization Authority (may be vague)
NVMA - Network Virtualization Mapping Authority (less vague)


BGP might not be a mapping service; but with a stretch of the imagination the 
mapping of NLRI to prefix/communities/etc could fulfil the litmus test.

The below words of 'server', 'controller', and 'orchestrator' carry a lot of 
baggage/meaning at the moment and might convey other meanings. I like the NVMA 
as a functional element.

NVMA +1

Truman

On 9 Apr, 2013, at 9:04 PM, Lou Berger 
<[email protected]<mailto:[email protected]>> wrote:


Also...

NVS - Network Virtualization Server
NVC - Network Virtualization Controller
NVO - Network Virtualization Orchestrator

On 4/9/2013 8:27 PM, Larry Kreeger (kreeger) wrote:

Lou,

Given your criteria for starting with NV, we could potentially have one of
the following:

NVS - Network Virtualization Service (service seems vague and perhaps an
overloaded term)
NVA - Network Virtualization Authority (may be vague)
NVMA - Network Virtualization Mapping Authority (less vague)
NVO - Network Virtualization Oracle (assuming oracle is OK when it has the
NV prefix)

- Larry

On 4/9/13 4:51 PM, "Lou Berger" <[email protected]<mailto:[email protected]>> 
wrote:


Larry,
           I think the NVC and NVO terms are fairly solution neutral, i.e., I
don't see why not.

I guess your question does lead to yet another option
NVS: Network Virtualization Service

I personally like NVC, but given that we have NVEs, I'd be happy with
anything of the form "Network Virtualization <foo>" as long as <foo>
doesn't start with an "E".

Lou

On 4/9/2013 7:25 PM, Larry Kreeger (kreeger) wrote:

Lou,

Do you see the terms "Controller" or "Orchestrator" standing up to David
Black's "BGP" litmus test?

- Larry

On 4/9/13 3:18 PM, "Lou Berger" <[email protected]<mailto:[email protected]>> 
wrote:



On 4/8/2013 9:12 PM, Larry Kreeger (kreeger) wrote:

At the NVO3 WG meeting in Orlando, I brought up some suggestions for
terminology changes/additions and had an AI to bring this to the
listŠso
here goes.  I would prefer to see constructive suggestions in
responses.
In other words, please suggest something better if you object to
these.

1) Oracle -> Information Mapping Authority

Stewart mentioned that he has copyright concerns with using the term
"oracle", and others have expressed distaste as well.
In draft-kreeger-nvo3-hypervisor-nve-cp-01 we replaced the term with
"Information Mapping Authority" (IMA).  We would like to get consensus
on using this new term in all WG documents going forward.  In the
meeting Linda  expressed a concern that IMA might get confused with
IMA
being confused with the acronym for Inverse Multiplexing for ATM, and
suggested something like "Directory Service" to which David Black
replied that she might have trouble convincing people that BGP can be
categorized as a "Directory Service".

Perhaps we're over thinking this one.  We have NVE, so why not
NV Controller (NVC) or NV Orchestrator (NVO)?

Lou


...





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



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

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