IPv6 deployment scenarios – Brian Carpenter
- Assumed v6 deployed by v4 ran out
- change transition model
- More internetworking than original ipv6 design originally anticipated. Assume v6 clients will need to access v4 servers forever
- Tunnels – Dual Stack Lite ( share ipv4 addr amung custs by combining UPv4-in-IPv6 and NAT, Driven by Comcast BB model ) – 6rd ( blend of 6to4 and ISTAP providing atumatic tunning of IPv6-in-IPv4 to ISP subscribers. Deployed by Freenet.fr)
- Older mesh and hub+spoke models also documented.
- NAT64 – old NAT-PT deprecated
- NAT64 – millions of IPv6-only custs needing access to IPv4-only services
- NAT64 only solves 1 problem – cannot be met my dual-stack – DNS64 dns server creates AAAA of site only with A record. Packets to NAT64 box and translated
- Various problems. 7 ietf drafts. Only solving since case
- V6OPS WG- Emerging Service Provider Scenarios for IPv6 Deployment – ID and survey ISPs then publish draft 03/2010
Rapid IPv6 Deployment in ISp Network – Skeeve Stevens
- AIM – Get people to use IPv6
- eintellego runs ISPs
- What stopping ISps implimenting IPv6
- Why not? – Too expensive , bigger ISPs yes, smaller ISPs perhaps not, NOT expensive to do enough to be able to play with it
- Why not? – Too Hard – Lack of internal skills – IPv6 is NOT hard, cisco admin should be basic IPv6 in 2h and IPv6 BGP in under a day – Play now or else you will be overwhelmed later when everybody is yelling
- Why Not? – Don’t know where to start – Start with a external co-lo box in the US – Allocate small amount of time – Get access to a lab – Start at the border
- Why Not? – No one asking for it – True enough – Don’t know about Ipv4 exhaustion, but they will
- Why Not? – Little vendor support – improving – DSL CPE equipment getting better – Carrier Grade NAT ( CGN/LSN)
- Why Not? – What is IPv6? – From Many IT professionals – Integrators have minimal experience
- Why Not? – Who can help me? – commerially, very few people – Some training courses – Community helps
- IPv6 is big, break it down into stages
- Experiment Externally
- Get allocation from APNIC
- Enable your Edge (BGP)
- Enable Core
- Enable desktop
- Enable your hosting
- Enable Operation Support Systems
- One hosting company just took 1 week
- Very rapid training, just a couple of days
- Simplified addressing – short to medium term – rapid deployment – format – 2406:9800::F:203.18.102.99 – Use F0 instead of”F” for next pop – Using /128s will increase routing table – “chazwazza” is ipv6 equiv of “octet”
- We use /64 for all end customer assignments – static routes to make v4-in-v6 work
- NTP might not work
- Some security concerns
- Go through commons OS, Daemons, Hardware ( phones, printers, UPS, gameboys)
- Might have to tunnel
- Hassel carrier if not provided
- Hassel vendors if they don’t work
- Some parts won’t happen overnight
- Predictions – Telstra selling IPv6 mid 2010 – Resource rush to grab IPv4 IPs while they can , surge in APNIC membership – exhaustion brought forward – secondary market will come – APNIC will lose control
Simply allocation of ipv6 addr to ipv4 holders – Elly Tawhai
- Policy 73
- Encourage greater uptake of IPv6
- An APNIC member with IPv4 allocation is eligible /32 . Member with assignment gets a /48
- One-Click IPv6 from my.apnic.net
NZ/IPv6 from (offshore) DNS – GGM (no name)
- Passive tap on DNS servers – spot reverse lookups for in-addr.arpa
- Capture all DNS in 1 day look for NZ IPs
- 1 in 10,000 lookups are doing IPv6
- 1 in 200 queries for DNS using IPv6
- 87.5% active delegattions in 24 hour period
- 45% of V6 networks live in 24 period
- 52% of v6 is Macs
- IPv6 not on the phone
- 6to4 common even with providers that do IPv6 native
Things running late so IPv6 panel skipped.
