Recently, there has been a discussion on what to use for point-to-point links under ipv6. Over the years I have seen providers use a wide variety of subnets across a point-to-point link. Anything from a /64, to /122,124, and the 127. There are arguments for each. But let’s look at some RFCs.
This was the original RFC on point-to-point links advocating for a /127. It was moved to a “historical” status in 2012 after a /127 was found to be damaging.
This RFC explains some of the issues with using a /127. In a nutshell.
Using /127 can be especially harmful on a point-to-point link when Subnet-router anycast address is implemented.
All of this brings us to some current RFC drafts and further reading
Some notes on point-to-point addressing
1. A subnet mask using something shorter than a /64 breaks some IPv6 functionality. A point-to-point link does not use the “broken” features anyway.
- The above 6164 RFC said vendors had to support /127s. Many wrote code to comply with the RFC, which has now been obsoleted.
- A /64 could expose the link to security issues.
As of this writing, there are many approaches. One approach is to set aside a /64 for the point-to-point but only use a /127 out of that /64. You don’t re-use anything else out of that /64. Other approaches involve using a /126, /120 or a /112 are being accepted until this is all figured out. So, why not a /122 or something? In short, it all has to do with the math of the subnet breakdown.