[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [narten@us.ibm.com: PI addressing in IPv6 advances in ARIN]

On Apr 16, 2006, at 12:49 PM, Jeroen Massar wrote:


[very nice cross posting going on here ;) ]
Nah, I just hit "reply-all", but only one actually made it through. I'm not subscribed to the rest of the lists. 

I've lowered the CC list to something more reasonable now.
Oh, and commenting derisively on something you do yourself seems a bit silly. 



On Sun, 2006-04-16 at 12:10 -0400, Patrick W. Gilmore wrote:
[...
large snip about trying to bash shim6 which is not finalized
yet, thus how can you bash it ?
Note: extra sarcasm included in this post. Eat the eggs with salt.
...]
I don't remember bashing shim6. I remember saying people do not agree it is the way to go.
As for "finalized", if I don't agree with the basic idea of a technology (e.g. inserting a "shim" into the IP packet), how can you "finalize" it to something with which I will agree? 



Oh, and one thing I should have said last time: Technical arguments
are important, but they are only part of the decision process.


In other words: "You are right with your arguments, but I just threw
your args away as they are futile based on the comparison of money
earned this way or the other."...
I'm going to assume you are being sarcastic here, since your "translation" is factually incorrect. I was clear the technical arguments are not sufficient, or even close. I was adding that the business elements are an additional hurdle.
BTW: Sarcasm is usually intended to either be funny or illustrate a point. Your sarcasms is definitely not funny, and the only point you are illustrating here is a complete misunderstanding of the discussion at hand. 



People (like me) have explained that the Internet is a business, and
in addition to being .. technically unsavory to many people, shim6 is
simply not viable in a business setting.


And as you will only care for your business for the coming 10 or maybe
20 years you really can't care what happens to the internet afterward.

The idea of IPv6 is (still not was) to have it around for quite some
time longer than the lifespan of IPv4. Fortunately, the PI thing is far 
from the end of the world and will only help catch on, see below.

Of course any vendor will love the idea of having to do another IP
version of course, bring in the cash ;)


This is close to a useful argument.
First, predicting things like router-engine capabilities 20 years in the future is beyond silly.
Second, the is a very real possibility that arguments about 'blowing up the routing table' are completely incorrect. People have been worried about it for over a decade and it has yet to come to pass. (You could argue that it hasn't happened because people are worried about it, but that it true with or without shim6, so it's a wash either way.)
Lastly, whether it is right or wrong, getting businesses to do something based on a 20 year horizon, especially when it is painful today - and will be for the next 20 years! - is essentially impossible. So why are you trying to get them to do it? Personally, I have much more important windmills at which to tilt. 



Neither backbone operators
(vendors) nor end users (customers) are warming to the idea.  Just
the opposite.  (At least in general, the one-in-a-million end user
with DSL and cable who likes the idea 'cause he can't figure out how
to spell "B-G-P" or doesn't want to pay for it is irrelevant.)


Irrelevant for you as they don't give you money. Indeed, you only look
at your own business interrest (and who can blame you for that ;)
(Once though the internet was there for the masses and not only for the 
ones with cash)
No, irrelevant PERIOD. You cannot architect the Internet for the one- in-a-million end user, _especially_ one who does not pay for the infrastructure.
If you argue that they are at all relevant, then we have a lot more problems with shim6 than we've discussed. (And with the Internet in general.) So please explain to me why they are relevant in any way whatsoever? I am honestly eager to hear your thinking along this line.
Just be completely clear on the implications of "proving" a non- paying one-in-a-million end-user is reason enough to change the core architecture of the whole Internet. 



So how do you get a technology widely accepted when the majority of
people involved do not think it is the best technical solution?  When
the majority of vendors supposed to implement it will not do so for
technical -and- business reasons.


There is for you indeed a business reason to not like it: the end-site
won't have any reason to stick to the upstream. Which is indeed a bad
business for many of the 'vendors' you mean.
As Eliot Lear also said very clearly: Thanks for lining the vendors and 
all the stockholders pockets ;)
I'm not a backbone. I am personally an end user. And my company is not a backbone, and does not sell transit to anyone. In fact, we are probably the largest "end-site" consumer of bandwidth in the world. And I still dislike shim6 both technically and commercially, personally and professionally. So does every technical person at my company who has any interest in this topic. 

It is not just backbones.  Shim6 is not commercially viable.  Period.

But thank you for attempting to divert the real discussion.
That is in the long run, most likely in the coming 10-20 years the IPv6 
routing tables will not have 'exploded' yet, but the folks selling
equipment and having stocks of those venders after that most likely will 
have a nice retirement fund. Thanks to you!


First, thank you for thinking I am so important.
Second: Whatever. If you honestly believe cisco & juniper will fail or succeed based on shim6, you really need to reevaluate your hypothesis. 



Nevertheless, the PI thing is really *not* a bad thing, as it can be
used as an identifier for shim6, which is actually perfect. It just
saves on having to do a complete policy process for getting address
space for this type of usage. But thanks to this, this won't be needed
and thus in the end anybody who can get PI can use a shim6-alike
solution and won't have any problem with the upstream that actually
wanted to lock them in by letting them pay loads for an entry in the BGP 
tables.
Thus people voting for PI, thanks for helping shim6 or another solution 
in that space, progress a lot :)


Then why are we arguing about this?

--
TTFN,
patrick