One of my clients recently requested to deploy IPv6 access on multiple sites. There was a previous /28 assignment from ARIN that was partially deployed. I did follow some of the previous information to model my deployment but still wanted to spend some time reviewing ipv6 [RFC2460] and learn about the why and how to deploy the protocol. I reviewed some material from the CCIE books as well as attended a Webinar sponsored by ARIN which was on IPv6 Fundamentals.

Originated in RFC 2460 and replaced by RFC 8200 The IPv6 protocol has five improvements 1. The increased address space 2. Header Simplification 3.Improved Extension Support 4.Flow Labeling 5. Extensions to support confidentiality authentication and integrity. Address Types.

Ethan Banks [Twitter @ecbanks] tweeted about the ARIN sponsored webinar for IPv6 fundamentals. The first in a series of lectures led by Nalini Elkins. She has a remarkable credentials and has two RFCs. The first hour covered the IPv6 address structure. The 128 bit address length , represented in Hexadecimal. The various address types Global[2000::3] , Private/Link Local[FE80::], ULA(Unique Local Unicast)[FC00::7] , Multicast [FF]. The second hour covered ICMPv6 , SLAAC , Multicast Listener Discovery , and Router Advertisement for IPv6. I took plenty of notes from this course and am looking forward to the next webinar on May 7th History of IPv6: Past, Present, and Future , which is lead by Bob Hinden. Bob Hinden was a co-inventor of the IPv6 Protocol.

After a review and the webinar. I felt empowered to build the new IPv6 deployments. I started by assigning address space to each site. A /32 global prefixes to each of the sites. Broken into two /48 prefixes [XXXX:AAAA:0::/48 , XXXX:AAAA:1::/48]. One for internal use , another for customer assignment. Each 48 bit network prefix is followed by a 16bit subnet id, followed by the host. This allows the assigned of /64 on each interface/ customer. It seems like a tremendous waste of address space on a point to point link but it helps keep the address hierarchy.

OSPFv3 was deployed to support internal address routing. OSPFv3 uses the links themselves instead of the actual subnet for advertisement. This is due in part that interfaces typically have multiple IPv6 addresses and OSPFv3 uses the link local address space for communication. BGP was turned up both internal / external. The most interesting issue was creating a transit access list for IPv6 Bogon/Martial space.

My next phase of this project will be for automatic address assignment for the space and some additional learning on the protocol. I look forward to continuing making use of IPv6 more.

References: