I’ve been doing a lot of research into IPv6 deployment scenarios as more and more of my service provider customers begin looking for solutions. The main problem that needs immediate attention is the lack of IPv4 address space left for service providers to use for turning-up new customers.
Let’s start off with the basics. First, you’ll need to request an IPv6 block from ARIN (in North America). You’ll then need to identify if your current upstream provider will do IPv6 MP-eBGP peering with you. This is key. If they won’t, find one that will.
Now that you have an address block and its advertised out to the IPv6 Internet you have to create a migration strategy. Unfortunately this is where it gets tricky. You can’t just move all of your customers to IPv6 and be done. There are two main reasons for this:
- Not all websites (or other services) your users will access have an IPv6 equivalent.
- Not all of your users have devices (routers, PCs, etc.) that support IPv6.
- Run dual-stack v6 and v4 and use RFC1918 private addresses + NAT for v4
- Give all customers only a v6 address and use NAT-PT or NAT64/DNS64 (preferred) to translate for access to v4 sites
- Customer devices that don’t support v6 will need to be upgraded.
- Suffers from some of the same issues v4 NAT has