internal routing-protocols for IPv6

[Kristoff Bonne]

Greetings,

>> => RIPng is RIPv2 with IPv6 support. Cheap but very limited...
> True, but I guess this is a first step.
> At least, it's better than static routing. ;-)
>> => it is why RIPng is heavily used as an IGP today (for instance I use
>> it here). As far as you know the limits it is a good choice.
(...)
>> => iBGP is a weak part of BGP, for instance the full mesh constraint is
>> a real pain (and confederation/reflectors nighmares).


To start, I was thinking of the following topology:
- two 'backbone-node' where we pull in the tunnels from the 6bone. (two
nodes, for redundancy)
- From those 'backbone-nodes', tunnels towards the 'internal' networks.

I would like to use both cisco-routers and unix-routers for this.

So, I guess I have two options: RIPng and BGP.
(I'll try both to see what turns out to be best).


>> If I remember correctly from the time I used this), one of the great
>> things about ISIS, is that you can use a single routing-protocol to
>> carry both OSI and IPv4 routing.
> => this is the Ships-In-the-Night argument.
Euh ... what is a 'Ships-In-the-Night argument' ??? (what does this mean?)

>>>> - internal BGP?
>>> => *not* an IGP!
>> Technically speaking not, but you could use it as a IGP (just assign a
>>   private AS-number to your 'customers').
> => iBGP is a weak part of BGP, for instance the full mesh constraint is
> a real pain (and confederation/reflectors nighmares).

Well, I would use my backbone-AS on the 'backbone-nodes', but private
ASnumber on the 'internal networks'. In that case, there is no need for a
full-BGP-mesh.

Cheerio! Kr. Bonne.
-- 
KB905-RIPE (HOME)                               belgacom internet backbone
(c=be,a=rtt,p=belgacomgroup,s=Bonne,g=Kristoff) International Connectivity
kristoff@belbone.net                            fax: +32 2 2435122