Re: IPv6 multihoming

[Vlad Ion <vlad.thoth@gmail.com>]

Hi Janos,

Its simple:

1&2. Make the whole 2002:: class provider independent, not provider aggregable so it will be dedicated for transition and multi-homing purposes only
.
3. Just allow everybody to use their IPv4 address as PI IPv6 address. I see no problem with that. It even makes it a lot easier for everyone to transition their existing space to IPv6.

4. Everybody should accept /48 in IPv6 routing just because there's a clear need to have multi-homing and you can't get around it. A simple procedure to reduce, in time, the size of the /48 classes that are announced is for people to slowly give back their v4 space to the RIRs as v6 expansion gains momentum and slowly dedicate v4 and their 2002:: counterparts to multi-homing only.

5. All ISPs and carriers that take the possible RIR offer to take free IPv6 space in exchange to announcing their v4 space in v6 format should deploy 6to4 relays. Relays should also be deployed by anyone wanting to make use of v6 multi-homing.

I agree, 6to4 is a good solution, but the problem remains with the slow deployment rate of IPv6. The problems IPv6 has with multihoming were the key factor in all the deployment projects I saw that gave up eventually on IPv6 be it in telco or enterprise scenarios.

One fact remains clear as daylight: a unified solution has to be picked that solves the multi-homing issue and a strategy has to be adopted to provide access to the existing v4 internet in v6 form before we run out of v4 space and more and more 2-layer deploayments of NAT will be introduced such as carrier-grade NAT


Best regards,
Vlad Ion.

On Mon, Jan 25, 2010 at 12:48 PM, Mohacsi Janos <mohacsi@niif.hu> wrote:

On Mon, 25 Jan 2010, Vlad Ion wrote:

Hi,

For a year now whenever it comes to IPv6 telco implementations I keep facing 2 problems so I was hoping you can
guide me towards find a group that deals with these issues or discussion solutions. The 2 problems are related to
IPv6 multi-homing and access to the internet in IPv6 format for a quick transitions from v4 to v6. Also I need some
guidance as to what needs to be done for a draft document proposal to be created about the proposed solutions
mentioned bellow and who needs to be involved in this process.

As far as multi-homing goes in IPv6 the solution discussion generated by using provider-independent address space
like mentioned in draft-hain-ipv6-pi-addr-10 seems too complicated to implement efficiently and generates a lot of
unnecessary work. Because IPv6 will never really be adopted by ISPs, telco and enterprises until it offers a
feasible multi-homing solution my proposal is that some solutions are redefined such as provider independent address
space and the 6to4 standard.

I propose that the 6to4 ip conversion space from ipv4 addresses to 2002::ipv6 space will be redefined as provider
independent address space. This way whoever wants to implement ipv6 with multi-homing can simply redefine their
existing IPv4 addresses in IPv6 6to4 format and have multi-homing in ipv6. Everyone already uses ipv4 multi-homing
with success so I see no point in defining a new addressing system for v6 when everyone can simply use the same v4
address space for multi-homing but converted in 6to4 format.

I believe this more a policy and RIR business to treat 2002:pi_ipv4 as PI IPV6. May I have some questions:

1. Do you want to treat differently 2002:pi_ipv4 and 2002:pa_ipv4? 2. How do you recognise that a particular IPv4 address PI or PA?

3. How do you prevent everybody to use their IPv4 address as PI IPv6 address?

4. What will happen if everybody will use their single /32 IPv4 address as IPv6 PI address space?  Everybody should accept /48 in IPv6 routing?

5. Who will operate 6to4 relays? Every Tier1? Every ISP? Every PI customer? Everybody?

Also, another issue faced by whoever uses IPv6 is that access to the internet in v6 format is limited so a proposal
has to be made to the RIRs to offer incentives such as free IPv6 space for anyone who implements 6to4 relay routers
and advertises their existing v4 space in v6 format along with the newly received free v6 space.

6to4 is very good solution for interconnect IPv6 islands, but fundamentally relying on existing IPv4 infrastructure. In long term the IPv4 usage should be decreased as IPv6 fully adopted.

I believe that as long as ietf gets involved and a rfc is written on these 2 proposals starting with the redefining
of  the provider independent address space and its inclusion in the 6to4 format things will be a lot more compact
and give some additional momentum to the IPv6 migration process.

Best Regards,
               Janos Mohacsi