[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: [BEHAVE] [Softwires] [Int-area] next steps for ipv4-ipv6 co-existenceandan interimmeeting



 

>-----Original Message-----
>From: Mark Townsley [mailto:townsley@cisco.com] 
>Sent: Monday, August 25, 2008 7:01 AM
>To: Dan Wing
>Cc: Templin, Fred L; 'Jari Arkko'; 'Internet Area'; 
>behave@ietf.org; 'IPv6 Operations'; softwires@ietf.org
>Subject: Re: [BEHAVE] [Softwires] [Int-area] next steps for 
>ipv4-ipv6 co-existenceandan interimmeeting
>
>Dan Wing wrote:
>>>> Did I miss anything?
>>>>       
>>> I was not able to attend the meeting; were considerations
>>> for ipv4-ipv6 coexistence in enterprise networks discussed?
>>>     
>>
>> I believe most everyone is thinking of the Internet, rather
>> than enterprise networks.
>>
>> If there are unique enterprise requirements it would be 
>> valuable to incorporate them into a requirements document 
>> (either in draft-ietf-v6ops-nat64-pb-statement-req or in the 
>> document that Jari's email said that Jari and Mark Townsley 
>> are writing -- I'm not sure which one would be best).
>>   
>Enterprise networks did come up, at least twice, at the Mic.
>
>First, I don't think we need to reiterate any enterprise requirements 
>that are solved by dual-stack deployment. In the vast number of cases, 
>this is the most general and safest bet to begin IPv6 
>deployment today. 
>The real argument is where the tipping point is for when one can and 
>should turn off IPv4 within their network, and whether deploying 
>additional tools to help accelerate this are on the whole advantageous 
>or more trouble than they are worth.

IMHO, IPv4 makes a fine link layer for IPv6 and based on
what I can tell this is especially true when there may be
competing interests within the enterprise such that it is
not just "one big happy family". Also when there may be a
need for deeply-nested enterprises-within-enterprises.

>Some passionate arguments 
>were made 
>that, indeed, there are significant operational advantages to 
>moving the 
>network away from dual-stack and directly to a single IPv6 stack model 
>in short order. Further, that in settings where the network 
>equipment is 
>relatively modern and the applications fairly well-known or 
>under tight 
>control, that it is conceivably a realistic option even today (or at 
>least "real soon now"). At this high level, I think that the 
>case falls 
>roughly into 1.d, i.e. something similar to what I had depicted as the 
>"greenfield wireless network." Tools that could be useful here include 
>some version of dual-stack light or NAT64, depending on how 
>much control 
>you have over your host stacks.

AFAICT, dual-stack light/NAT64 approaches would work fine
in a coexisting IPv6/IPv4 network. It may look like NATs
within NATs at the IPv4 level, but the IPv6 overlay sees
it as all-IPv6 with no NAT traversal requirements. 

>Which to deploy really depends on the 
>pain threshold of obtaining equipment for and running the two IP's 
>side-by-side, on top of one another, or translated at an edge point.

Or all of the above.

Fred
fred.l.templin@boeing.com

>
>- Mark
>
>> -d
>>
>> _______________________________________________
>> Behave mailing list
>> Behave@ietf.org
>> https://www.ietf.org/mailman/listinfo/behave
>>
>>   
>
>