(6bone) bunch'o topics

[Michel Py]

[consolidated multiple posts]

> Ben Winslow wrote:
> I wish it weren't the case, but there are still cases where you
> have to accept the 'best you can get' scenario.  I do agree that
> as much effort as possible should be taken to get a good peering
> point (it's my opinion that latency is a huge problem on 6bone
> right now--one of the joys of having an almost guarenteed 10 IPv4
> routers between every IPv6 hop), a slow peer is better than no
> peering at all.

I agree with Ben here, not to mention that getting it for free is not
bad either.


> Pekka Savola wrote:
> Sometimes IPv6 is actually used in e.g. SSH sessions, WWW-pages etc.

Sometimes. And I'd rather waste hours of configuration for the sole
purpose of telling my email buddies that I can play Quake on a v6-only
server than look at a meaningless web page that says "I have a v6-only
web site". I remember ten years ago configuring bridging of IPX over
routed IP to play doom (it was IPX broadcast based, therefore the
bridging).


It appears to me that some of you are missing the point:
- Viagenie is the biggest IPv6 ISP today. It makes sense to have a
tunnel with them anyway (especially for free) and that is why I have
one.
- It is legitimate for Viagenie and any other large ISPs to service
customers in the entire world. Since we do not allow them to advertise
regional prefixes but only their pTLA /24, long RTTs are to be expected.
- Advertising regional prefixes would be useful if they had a native
backbone, and it's not with the money we are giving them (none) that
they are going to build it.
- I would pay for native v6 service, if I could get it that is, but this
means native all the way to a major v6 backbone, which again does not
exist as we speak. I don't call one ISP providing native v6 to the hotel
in Minneapolis a backbone.


> Scott Martin wrote:
> I suppose that until ipv6 can make a corporation money, or until the
> govt steps in with large amounts of $$ for funding, this "project"
> will still flounder about... Sorta like the mbone.

Right on the money (pun intended).

> Personally, I would love to be running native ipv6 - no
> renumbering worries to deal with.

If you could get addresses of your own and these don't exist either.


> Bill Sommerfeld wrote:
> I'll point out that given the prefix length restrictions on the
> 6bone, getting a pTLA is just about the only way for 6bone sites
> to set up redundant routing.

For the same matter, getting a subTLA too. This is not new, has been
posted before and is a direct consequence of the development of v6
multihoming being torpedoed.


>> The 6bone is _not_ real world. There is no real world, as of today.

> Joe St Sauver wrote:
> Sure there is. In fact, I would assert that IPv6 is very real real
> today, and v6 is poised for accelerated availability (at least in
> the research and education network community)

The real world is when I can get to a multihomed IPv6 www.cnn.com,
www.ebay.com, www.etrade.com, www.cisco.com and microsoft.com over
native v6. Today, v6 is a toy and if was not for v4 most of us would not
be able to read this. Lots of us here are working to make the toy
usable, but it does not change the fact that is still is a toy and will
remain a toy until we provide:
a) Multihoming.
b) Geographical addresses, preferably PI.
c) Provider-independent (PI) addresses for large organizations or big
content providers.

Until these three issues are resolved (they are for v4) who can even
pretend to be "serious" about IPv6? A serious global ISP that aggregates
its entire worldwide address space into a single prefix? A serious
portal using their ISP's address? A serious mission-critical singlehomed
site?

The problem is not latency over 6bone tunnels. The problem is that
nobody is willing to fork out the cash to build a real v6 backbone
because there is no revenue on the horizon to pay it back, because major
issues have not been addressed.

Michel.