Re: [v6ops] I-D Action: draft-ietf-v6ops-ipv6rtr-reqs-01.txt

[Fred Baker <[e-mail]>]

On Jan 2, 2018, at 10:52 AM, [e-mail] wrote:
>> A New Internet-Draft is available from the on-line Internet-Drafts directories.
>> This draft is a work item of the IPv6 Operations WG of the IETF.
>> 
>>        Title           : Requirements for IPv6 Routers
> 
> I've addressed all of the comments I had notes for except --

Thanks for this. You will have seen, by now, my request for online discussion of the draft.

If I have one comment, it would be this:

The document is developed in comparison to RFC 1812, which I know a little about. RFCs 1122, 1123, and 1812 were developed by the community (I came in late in the game) to update and address a variety of small issues that had been identified and sorted out. Today, if we were doing this, we might start with an "erratum", such as "let's clarify: subnet broadcast address has a host part that is all ones, not all zeroes", and we would address the issue in some subsequent update to the relevant RFC. In the late 1980's, there was a perception that "OMG, updating each RFC would be a huge job". I think Bob Braden might agree with me when I say "updating them all in a single omnibus document, so big it became three, was a huge job".

The major advantage that RFC 1812 had was that it was edited when CIDR was being agreed to and worked on in a number of working groups, and we could address CIDR in one broad document. It was also at a time when the behavior of several routing protocols (RIPv1, which made non-CIDR assumptions, OSPF with TOS-Based-Routing, and IS-IS with TOS-Base-Routing) needed to be described so that networks could deploy them interoperably. The practical objective was to "finish" work that had been going on for some time and lost its way; the theoretical objective (stated with the benefit of 20/20 hindsight) was to describe a new commercial Internet that was just coming into being.

The huge benefit that RFCs 1122, 1123, and 1812 brought to the community was not WHAT they indicated had been decided. It was WHY, among several possible solutions, they had been selected. The three documents gave vendors, open source, and other developers a view into operational experience regarding what worked, what didn't, and WHY some solutions were better than others.

If there is one thing I want to find in this document, it is WHY, backed by operational experience.

Let me put this in a specific request to the authors. I would like to hear from multiple networks - LinkedIn, yes and of course, but also others - what you would like to mention by reference in your RFPs. How do you intend to build your networks? What requirements does that impose on each bit of equipment?

As a transition mechanism, I can imagine that you have micro-services that are IPv4-only, at your ingress that it might be load balancers serving a web-based service, and in the back end it might be "simple" translation such as proposed in IVI or SIIT-DC. Your network therefore looks something like:

       ,-.               ,---.                ,---.
      /   \             /     \              /     \
     /     \          ,'       `.          ,'       `.
    /       \        ;           :        ;           :
   ;         :       ;           :        ;           :
   ;         :      ;             :      ;             :
  ;           :     ;             :      ;             :
  |           |    ;               :    ;               :
  | The Great +----+   IPv6-only   +----+  IPv4-only    |
  | Unwashed  |XLAT|  Data Center  |XLAT|  Back End     |
  |           +----+               +----+ Microservices |
  |           |    :               ;    :               ;
  :           ;     :             ;      :             ;
   :         ;      :             ;      :             ;
   :         ;       :           ;        :           ;
    \       /        :           ;        :           ;
     \     /          `.       ,'          `.       ,'
      \   /             \     /              \     /
       `-'               `---'                `---'

Tell me what you require of those equipments, and more specifically, why? Is HTML the waist of your hourglass? Tell me about that?