Saving IPv4 Addresses on Interconnection with RFC8950
• 6 min read
Many of us have been struggling with dual-stack IP address assignment on transfer networks. But what if we could reduce the work and concentrate on one IPv6 address per interconnection? Turns out, we already can. This article sheds some light on the practical implementation of RFC8950 at IXPs.
“It’s a practical idea, for sure I’m going to test this today on my Nokia 7220 spine/leaf lab running Spine & Leaf fabric EVPN/VXLAN. thanks for sharing and pointing something very helpful. it has practical applications in advance DC's.”
Indeed, DC fabrics were the first large scale use of BGP Extended Next Hop. Together with Link Local addressing and BGP unnumbered RFC8950 drastically reduces IP addressing overhead and simplifies setup. One can even see that FRR is biased by this use-case (by default FRR sends Router Advertisements upon enabling extended next hop). Now that Extended Next Hops are in broad use in DC fabrics, it's time to exploit the same functionality for interconnection.
Showing 1 comment(s)