Re: [v6ops] Discussion focus: draft-ietf-v6ops-ipv6rtr-reqs

[Bob Hinden <[e-mail]>]

Tim,

> On Feb 6, 2018, at 1:41 AM, Tim Chown <[e-mail]> wrote:
> 
>> On 5 Feb 2018, at 21:52, Bob Hinden <[e-mail]> wrote:
>> 
>> Barbara,
>> 
>>> On Feb 5, 2018, at 9:10 AM, STARK, BARBARA H <[e-mail]> wrote:
>>> 
>>>> Thanks for this. It's cogent and to the point.
>>>> 
>>>> I do have a question. The authors are from LinkedIn and Comcast. How
>>>> would you interpret that fact in the context of the facts you point out?
>>> 
>>> That's for them to say what their motivation is. But Comcast and LinkedIn are both known to be at the forefront of IPv6 deployment. I wouldn't classify LinkedIn as an ISP of any sort or a core/transit network provider or an IXC. If LinkedIn or Comcast thinks there is a need for requirements for data center network routers, that would be interesting to know. Most major data center operators I know have gone to using SDN (with open source router software on commodity hardware). I don't recall having heard Comcast say they need a router requirements document to help with their access/regional network IPv6 deployment (which seems to be doing quite well). I don't know about any internal corporate networks they may have -- which I would classify as "enterprise networks".
>>> 
>>> Personally, I haven't heard any of the people working on any of my employer's access/regional/core networks asking for a new IETF router requirements document (the IPv6 metrics for wireline and wireless access are looking pretty good with non-IPv6 traffic due to hosts/wireless UEs/retail CE routers and not the network routers, and no issues in the core; router requirements would have no impact on use of 6rd in legacy DSL access). I have not yet noticed IPv6 in the corporate (enterprise) network that I attach to for work. I've also heard from some people who work with enterprise customers that adoption among those customers is a painfully slow crawl.
>> 
>> I agree with what you have written.  It’s not clear to me what the need is here.  Regarding enterprise deployment, I would be surprised if the reason for slow IPv6 adoption has anything to do with the IETF writing a document describing what an IPv6 enterprise router should implement.  If someone has evidence to show it is needed, I would like to see it too.
> 
> I certainly hear questions from campus network admins about IPv6 requirements for tenders. There are documents like RIPE-554 which are helpful in that regard (though it needs an update), but being able to point at an RFC has some value, and any update of 554 would likely draw on the consensus expressed therein, along with 6434-bis.
> 
> I really like the writing style of draft-ietf-v6ops-ipv6rtr-reqs; there's a lot of very good and very useful discussion and advice in there aside from specific MUSTs and SHOULDs.

I agree.  I was thinking that the draft would be better if it went through a list of features relevant for routers, discussed scenarios where they are useful, but didn’t try to get into MUST or SHOULDs because there are too many different types of routers.  For example, DHCPv6 prefix delegation is useful for ISP facing home/SHO routers.

Bob