[nycbug-talk] For and Against IPv6

[Isaac Levy]

On Oct 11, 2007, at 4:13 PM, Alex Pilosov wrote:
> On Thu, 11 Oct 2007, Isaac Levy wrote:
>>>> Not surprisingly, the sales guy didn't know anything about IPv6
>>>> addressing, and I explained the rammifications of these points:
>>>>
>>>> 1) IPv4 gives us 4billion+ addresses, vs. IPv6 and 32Undecillion+
>>>> addresses
>>> who cares as long as you can get ips
>>
>> My current DSL isp charges me $5/mo for each IP now, new policy in  
>> the
>> last year.
> it's pure profit, you understand that, right? arin charges like 1000 
> $/year
> for /20.

Does Pilosoft then not charge for additional IP's?  If so, can't get  
at least a /8 from you if I buy a Pilosoft DSL line, how about just  
a /24 for starters?

I'll use every single address immediately.

--
But aren't there costs managing the individual IPv4 addresses?  I ran  
a web-hosting ISP, you run a full-blown ISP- doesn't someone have to  
get paid to manage netblock usage, as a part of managing the  
service?  Doesn't that task become more time consuming (and  
difficult) as netblocks become more completely utilized?

>
>>> ipv6 is going to be a nightmare once it is actually deployed. i  
>>> don't
>>> know what gene was saying, but from operator's perspective, it isn't
>>> necessarily true.
>>
>> Um, wasn't there anyone screaming when IPv4 replaced the other
>> networking protocols of the day?  (which supported something like a
>> paltry 255 hosts across some networks?!) /me digs in the history  
>> books
>> for some info...
> ipv4 had 'flag day'. when the old protocol was disconnected and  
> everything
> was switched over to ip. clearly, the interwebs was smaller then.  
> [just a
> history lesson]

Awesome- can you point me at any historical references?  This is the  
fun stuff I love...

Are you referring to NCP transitioning?

>
>>>> 3) The rest of the developed (and much of the underdeveloped)   
>>>> world
>>>> is deploying it for production use, only Univ./Govt. are horsing
>>>> around with it in the US...
>>> not really no, nobody cares, except for japs.
>>
>> What about Russia, Alex?  I guess they have plenty of IP's.
> Until RIRs say "we cannot give you any more IPs", there's no  
> problem at
> all. Everyone can get IPs today. Thus, nobody feels the pain.

Again, where is my free /24 for my home?  I'll use every one, right now.

>
>> Have you or your customers done any business with Taiwan or China
>> recently?  My people have...
> What's that have to do with anything? are you saying that people in  
> taiwan
> have v6-only connectivity and unable to communicate to v4 sites?  
> [for the
> record, i do believe there are some providers who give by default  
> both v4
> and v6 ips. i don't know any provider in any country who'd give v6  
> only ip
> and 6to4 for the v4 access].

Gotcha- who are the providers who do both v4 and v6?

>
>> Respectfully questioning your experiences behind this logic:  
>> Seriously,
>> without meaning to sound contentious- how much time do you spend  
>> using
>> the IPv6 internet to know all of this?  I comprehend your serious
>> expertise and knowledge regarding the IP networking, and I always
>> appreciate your low-level knowledge of networks as you contribute  
>> to the
>> list- but seriously- how do you know so much about who uses the IPv6
>> net?  (How much do you know about what people *do* with your own
>> networks?)
> i'm active in the nanog community (which is the network operator  
> community
> - people who actually run the large networks, like l3, vzb, etc), and
> moderate the mailing list (yeah i know, alex as a moderator).
heh
> v6 has been
> a staple of conversations for past 6 years. yet, it remains in sort  
> of r&d
> phase - you can't buy v6 native transit from anyone. the best you  
> can do
> is v6 tunnels. fun thing is asking the v6 presenters at the conference
> 'so, can you sell me v6 transit' and getting giggles from them.

Is scenario perhaps influenced because you moderate the Nanog list?  :)

>
>> My case:
>> After getting involved with various international business situations
>> this year, I recognize that organizations I work with would benefit
>> from IPv6 connectivity.  I now have a business case behind my need,
>> and I don't necessarily care if you understand that, or believe that
>> it's 'valid'.
> show me the benefit.

End to end connectivity between machines, making more devices  
available over the internet.
(I have many small networked systems in my home office I'm dying to  
have online TODAY.)

Wouldn't you'd sell more bandwidth, if more people used more internet  
services and devices?

> show the business need. you say you want 65k IPs -
> what for?

Home Style Answer:
Easy, personally, counting on my hands (the geek view):

I just counted 19 computers of various shapes and sizes,  
architectures- in my home office, which I use and would quickly  
connect to the internet.  Of those 18 computers, 6 of them have 2 or  
more ethernet interfaces, and I count 9 wireless interfaces.  I'm  
doing dev. work using jails, and working on jailed system  
architecture.  My apartment alone currently has 3 active /24 subnets,  
all NAT'd behind 2 public IP addresses, and only a few machines have  
services I'd want to keep local-only...
The rest, are actually would be REALLY valuable to me if they were  
publicly available- (and people I'm working with for personal, work,  
and open-source dev. projects)  This is just the stuff that I see out  
and around my office, some of it is noisy old stuff and only gets  
turned on periodically.

Got all the numbers there?  Sounds like any of our geeky home/office  
arrangements, right?

Not to mention that I'd love some IP space to use IP telephony (and  
video stuff), as well as room to simply experiment with the network/ 
netblocks itself...
We all do this on our LAN's, but it's more valuable to me to get the  
systems online!

I don't know how many machines I've got running, and this isn't NASA  
or anything here- lots of crappy machines- in an apartment, being  
routed with PC gear, but I'm truly not kidding when I say I could use  
65k IP's.

--
Work Style Answer:

Here's my current business need (growing in recent months, not  
conjecture):
I work for people doing business in China, Taiwan, Japan- and they're  
spreading sales and operations into Europe.  I won't mention the  
client, but I'll say that moving files around between the businesses  
is a huge part of their income- loads of multimedia, various web-apps  
and file servers for industrial manufacture and multimedia  
production, etc... (this is why I work for them, managing and  
building applications and networks).

So, with that stated, their customers in Asia, (mostly other  
businesses, not direct consumers), are increasingly mentioning  
services on corporate IPv6 networks, and they maintain a few IPv4  
points of entry into their network which Americans/Westerners touch.   
The servers they maintain on the IPv4 internet are saturated.  With  
that, it would actually be FAR more convenient if we all could simply  
connect to their servers via IPv6, and provide them usable IPv6  
service as well.

There's loads of confusion about IT from my employers, because  
there's enough difficulty with spoken language barriers...

Another business client is starting to sell products across the  
internet in Japan, and gee-whiz, do I have to go to Japan and get a  
cabinet to make this happen?  (oh twist my arm, right...)

One other incentive for me right now- you know how much money I make  
on all this IPv6 related work?  Right now, ZERO.  I have no IP's or  
pipe to work with.
But out of a desire to grow, I'm willing to invest the time and  
resources to make more (and more interesting) work for myself...   
E.G. it's worth the expenditure, based on interest, and trust in  
future returns.

So is that enough of a business case for me ask my ISP's for a  
production-grade IPv6 internet connection?



>
> <snip>
>
>>> we'll sell ipv6 (native) once someone wants it. put a business case
>>> forward, you'll get transit in a week.
>>
>> Really?  Do you mean you'll stitch something together quickly for  
>> me, or
>> it'll be a real, usable, production-grade line?
> there *is no production-grade v6*. at best, think internet in late  
> 80s -
> stuff that engineers tinker with, which goes down every so often.  
> no large
> carriers have v6 native ports to customers. NONE. Not one. On  
> backbone, I
> think there's *maybe* one that have v6-native-capable backbone -  
> everyone
> else has "overlay" network or tunnels. There are two that will  
> actually
> sell you the "v6 tunnels" to their tunnel servers (one per city).  
> Everyone
> else simply doesn't care. there's only a limited amount of peering  
> (for
> example, as opposed to full mesh for tier1s, there are maybe a  
> couple of
> peering relationships, not in all geographic regions etc -  
> resulting in
> your packets likely traveling halfway across the country multiple  
> times).
>
> since big boys dont care, it is engineering-heavy providers that  
> try to
> make something sort of work. he is doing their piece. occaid is  
> providing
> "ghetto backbone" for ipv6-enabled customers. but it is really tunnels
> over v6 network which are held together by spit and glue, running  
> over PC
> routers...
>
> i'm just telling you 'how it really works'.

I appreciate this info Alex, it puts a lot into perspective for  
American networks.

>
>> Based on your attitude about IPv6, on list over the months, I'm not
>> certain that I'd want to trust this line...
> v6 is not production ready.

Neither is v4, it could be argued :)

>>
>>>
>>>> My ISP sales guy, however knowledgeable, is not very technical- but
>>>> he's a manager (read: decision maker)- and to my delight, he was
>>>> excited to get this request from me...
>>>>
>>>> What other info can I give to this Non-Technical ISP rep, (or  
>>>> any ISP
>>>> sales rep), to help build a business case in his organization, so
>>>> we can
>>>> buy real IPv6 lines in NYC sooner than later?!
>>> put a business case forward. say "there is an active demand for v6,
>>> if you
>>> offer v6 transit you'll instantly sell 500 T1s and 10000 dsl lines".
>>> except, of course, its not true.
>>
>> How do you know that nobody will buy it?  Your market research?
>> (again, not meaning to sound contentious- you DO know a hell of a lot
>> about the data/ISP biz' in NYC)  However, I don't see Pilosoft listed
>> when I search for 'IPv6 isp nyc':
> i don't see anyone listed. nobody cares, apparently.

It seems in America, no ISP seems to.

>
> <snip>
>
>> Hrm.  Is this because you don't provide IPv6 connectivity, and don't
>> really want to?
> we can. for us, its not so much of a big deal - our equipment  
> supports it,
> and we can probably interconnect with occaid for v6 "backbone", and I
> think one of our transits might do v6 as well. so the expensive is  
> X hours
> of engineering to make sure everything works as well as training my  
> noc
> guys to know what v6 is (this is much harder than engineering).

We're talking about separate business cases,

+ YOUR business case. (selling/providing IPv6)
+ MY business case. (acquiring/deploying IPv6)

--
What follows is meant with a great deal of respect, but none the  
less, it's a challenge:

Your business case (as the ISP) is a serious and challenge, but an  
old one- an upgrade.  We all know upgrades are painful transitions,  
so much respect is due any ISP.
However,
Yet I don't see anything but frustrated excuses for "why nobody  
cares", yet you (and most network operators) are resident experts on  
the topic(?).  How are ISP's and network operators so expert in  
something they don't care about enough to use, or deploy?
How do you know the challenges before you do it?  I'm not saying it's  
not a risk, it sure as hell is- but realistically, that's where the  
'business case' is.  Whoever takes the risk, makes the money- (or  
perhaps looses it).

I'm simply getting tired of getting more finger-pointing up the chain  
from ISP's.
Backbone providers are never going to initiate v6-native-capable  
links, they have no/little need- they don't make money off the  
endpoints en'-masse.

I'm interested in hearing from ISP's who say 'we did something, come  
buy some', even if it's just spit-and-glue tunnels...

That's pretty much all I have to say in this thread until some ISP  
makes a move, in the meantime, it's more relevant to setup the  
tunneled networks on an individual/organizational scale...

Rocket-
.ike