Networking

08:28 AM
Jeff Doyle
Jeff Doyle
Commentary
Connect Directly
LinkedIn
RSS
E-Mail
50%
50%
Repost This

The Fear And Loathing Of /64s On Point-To-Point Links

I discussed in a previous article the necessity of abandoning IPv4 thinking when creating IPv6 address designs, and how our deeply ingrained need to conserve addresses can muddle our thinking. Nowhere does this conservative aversion to address waste snarl at us as menacingly as when we consider – completely compliant with the recommendations of ARIN and other RIRs – assigning /64 subnets to point-to-point links.

I discussed in a previous article the necessity of abandoning IPv4 thinking when creating IPv6 address designs, and how our deeply ingrained need to conserve addresses can muddle our thinking. Nowhere does this conservative aversion to address waste snarl at us as menacingly as when we consider – completely compliant with the recommendations of ARIN and other RIRs – assigning /64 subnets to point-to-point links.

"You want me to allot a subnet with 18 million trillion addresses to a link that will only ever use two of them? Are you kidding me?" We know all the arguments for what we get in exchange for squander: Easier address management with one-size-fits-all subnets; simpler address interpretation; scaling; flexibility.

But still. Only using two addresses out of 18 million trillion? (Saying "million trillion" is a lot of fun if you imitate Carl Sagan’s voice.) Well, ask yourself when a /64 is acceptable.

Most people would say they can accept it on a regular LAN or VLAN segment. All righty then. To be fair, let’s take a really big LAN. Say, 5000 devices. Is a /64 acceptable there? Yes, you say? So we’re wasting (1.8 x 1019) – 5000 addresses instead of (1.8 x 1019) – 2 addresses. The difference between 5000 and 2 relative to 18 million trillion is miniscule. It diminishes to practically nothing. If it were any smaller it would be the amount I’m being paid to write this.

And yet a /64 on a LAN is acceptable and a /64 on a point-to-point link is not. IPv4 thinking can twist our reason. All of this does not mean there are not reasons to use a prefix other than /64 on point-to-point links – it only means address waste is not one of them. In fact, there are dueling RFCs on the topic.

RFC 3627 makes its case right in the title: "Use of /127 Prefix Length Between Routers Considered Harmful." The central argument in the document, however, is not as striking as the title suggests. Here it is:

  • When you use a /127 prefix on a point-to-point link, you have exactly two addresses available: PREFIX::0/127 and PREFIX::1/127. The problem the RFC cites is that the router being assigned PREFIX::1/127 might add the Subnet-Router Anycast address, which would be PREFIX::0/127. Then the router on the other end of the link, configured with PREFIX::0/127 will fail the Duplicate Address Detection test.

    Previous
    1 of 5
    Next
    Comment  | 
    Print  | 
    More Insights
  • Comments
    Newest First  |  Oldest First  |  Threaded View
    cef1
    50%
    50%
    cef1,
    User Rank: Apprentice
    2/27/2013 | 9:16:25 PM
    re: The Fear And Loathing Of /64s On Point-To-Point Links
    Great article Jeff!
    Here is my question: unlike IPv4, IGP in IPv6 can peer with link local address except for BGP (as per my testing-i may be wrong) and so the only place i am concerned about p2p links is the BGP links. All my IGP can work off the link local. And, if i am not using link-local for IGP peering i could potentially use an addressing scheme for my p2p links that i do not advertise to the world and hence use anything i like from a scheme that i can make my self like 1::1 etc. I am only advertising the Stub networks and the loopbacks for management. All the transit links i do not want to advertise hence just make something up--am i missing anything ? i am sure i am :)

    Thanks!
    DaveStubbs
    50%
    50%
    DaveStubbs,
    User Rank: Apprentice
    3/8/2012 | 2:18:19 PM
    re: The Fear And Loathing Of /64s On Point-To-Point Links
    Doesn't it seem a bit unbalanced? On the first page of the article you make the point of quoting RFC 3627 as an argument against /127 but then you fluff the whole issue when you argue against /126. It doesn't follow, logically. From reading the facts presented in the article, and ignoring your opinions, it seems that /126 is the BEST option!

    Why is it that any new technology that starts out with what appears to be "lots of space that will never all be used" always has wasteful allocations recommended, until people wake up years later and figure out they need to re-jig things. Reminds me how CIDR addressing came in as a "fix" years after people had been stuck with rigid Class A/B/C allocations on IPv4. And here we go again...
    yannis_n
    50%
    50%
    yannis_n,
    User Rank: Apprentice
    2/14/2012 | 2:03:05 PM
    re: The Fear And Loathing Of /64s On Point-To-Point Links
    nicely written and very informative but I have a question: what do you mean when you say "disable NDP on ethernet links"? Avoid autoconfiguration? Supress RAs?
    GBANSAL500
    50%
    50%
    GBANSAL500,
    User Rank: Apprentice
    2/10/2012 | 12:05:56 PM
    re: The Fear And Loathing Of /64s On Point-To-Point Links
    extremely informative and superbly explained !! thx
    jdoyle
    50%
    50%
    jdoyle,
    User Rank: Apprentice
    11/21/2011 | 8:00:34 PM
    re: The Fear And Loathing Of /64s On Point-To-Point Links
    Hi Mark,

    That's correct. If the main reason for using something other than a /64 is to save addresses or avoid the security issues of unused subnet addresses, then go all the way and use a /127. But in my opinion if you already are using /126 subnets, re-addressing to /127s is probably not worth the trouble.

    --Jeff
    More Blogs from Commentary
    Infrastructure Challenge: Build Your Community
    Network Computing provides the platform; help us make it your community.
    Edge Devices Are The Brains Of The Network
    In any type of network, the edge is where all the action takes place. Think of the edge as the brains of the network, while the core is just the dumb muscle.
    Fight Software Piracy With SaaS
    SaaS makes application deployment easy and effective. It could eliminate software piracy once and for all.
    SDN: Waiting For The Trickle-Down Effect
    Like server virtualization and 10 Gigabit Ethernet, SDN will eventually become a technology that small and midsized enterprises can use. But it's going to require some new packaging.
    IT Certification Exam Success In 4 Steps
    There are no shortcuts to obtaining passing scores, but focusing on key fundamentals of proper study and preparation will help you master the art of certification.
    Hot Topics
    6
    IT Certification Exam Success In 4 Steps
    Amy Arnold, CCNP/DP/Voice,  4/22/2014
    6
    Edge Devices Are The Brains Of The Network
    Orhan Ergun, Network Architect,  4/23/2014
    White Papers
    Register for Network Computing Newsletters
    Cartoon
    Current Issue
    2014 Private Cloud Survey
    2014 Private Cloud Survey
    Respondents are on a roll: 53% brought their private clouds from concept to production in less than one year, and 60% ­extend their clouds across multiple datacenters. But expertise is scarce, with 51% saying acquiring skilled employees is a roadblock.
    Video
    Slideshows
    Twitter Feed