[arin-discuss] Trying to Understand IPV6

Owen DeLong owen at delong.com
Mon Sep 13 15:30:15 EDT 2010


Scott,

Let's put the true numbers in a little bit of perspective...

First, when 4.2 billion addresses seemed like a lot, it was because the internet was not anticipated to be something everyone would use every day. It was expected to be a research community project used by a much smaller research community and if that had continued to be the case, 4.2 billion addresses would still seem like a lot.

Compared to 6,8 Billion people, however, 4.2 billion addresses (actually more like 3.2 billion unicast addresses) doesn't seem like much at all.

On the other hand... Consider that there are approximately 30,000 ISPs world-wide.

If we were to give every single ISP a pair of /28s (that's 32 of the current /32 allocations) in each of the 5 regions (AfriNIC, APNIC, ARIN, LACNIC, RIPE), we would still only consume the majority of the current single /12 allocated to each RIR. There would still be 506 /12s left in the current /3.

There were some very good reasons to give out IPv4 addresses like candy, and, having done so is one of the major contributing factors to the success and broad deployment of the internet. Is it remotely possible that in 50 or 60 years, we will still be using IPv6 addressing and that current liberal allocation efforts could prove to be unwise for that future? Actually, yes, it is possible.

However, I propose that we approach that logically.

For the time being, liberal allocation seems to have little consequence and provide the best opportunities for improving the user experience and the scalability of the network.

As such, let's go ahead and be liberal in our allocations and assignments. Give each end site no less than a /48 (even a residential customer). Let's try this experiment until we
have used up 20 /12s. If we get to the point where we have used 20 of the first 512 /12s in less than 30 years, then, we have plenty of time after that point to develop different
allocation policy for the remaining 492 /12s before we exhaust the first 1/8th of the address space.

OK?

Owen

On Sep 13, 2010, at 10:46 AM, Scott Morris wrote:

> There really isn't a great repository for things, at least that I've seen...  But keep in mind that the binary works the same in IPv6 as it does in IPv4.  :)
> 
> So when you look at a /21, do you look at it as 2000'ish IPv4 addresses?  (At which point a single /64 is more than enough)
> Or do you look at it as 8 x /24 networks (at which point a /61 is the technical equivalent)
> 
> /64 is used by many people as the de facto "network" addressing for any subnet because of all the magical EUI-64 addressing to work (e.g. less whiny customer calls = better).  But anyone doing DHCPv6 will quickly learn that a /64 is a SERIOUS amount of addresses than can be broken down internally to any variation they really feel like.  (/126 = /30, etc.)
> 
> So to best answer that questions, you have to know a little about your customer and how they'll use the addresses.  /56 seems to be the way many folks are going with things, but that's 256 /64 networks or the veritable "boatload" of addresses that will be wasted beyond belief.
> 
> Remember in the old days everyone gave out IPv4 addresses like candy.  Back then 4.2 billion addresses seemed like a lot...  Today, 340 undecillion addresses seems like a lot... Times change though!
>  
> 
> Scott Morris, CCIEx4 (R&S/ISP-Dial/Security/Service Provider) #4713,
> CCDE #2009::D, JNCIE-M #153, JNCIS-ER, CISSP, et al.
> CCSI #21903, JNCI-M, JNCI-ER
> swm at emanon.com
> 
> Knowledge is power.
> Power corrupts.
> Study hard and be Eeeeviiiil......
> 
> On 9/13/10 1:24 PM, Schnell, Darryl wrote:
>> 
>> Can anyone recommend a good IPV6 website for Beginners? I’ve read about eight web sites which say the same things and I feel like my head is going to explode. I guess the problem I’m having is trying to understand how an IPv4 CIDR notation translates in an IPv6 CIDR in order to fill out ARIN IPV6 Allocation Template future usage section. My actual question is –
>>  
>> If I assigned a customer say an IPV4 /21 in IPV6 this would translate into a /56? If I’m not mistaken a /56 would translate into something like 65,000 host addresses? That just seems like a lot of hosts to me, especially when most of the time I’m working with networks that are /26 or smaller. I guess my big problem is confusion over labeling. What would be the equivalent of a /26, /27, /28 or have we done away with blocks that small and simply would just assign a /56 instead?
>>  
>> Does any of the gibberish I wrote make any sense at all?
>>  
>> Any help anyone can offer is much appreciated.
>>  
>> D -
>>  
>> 
>> _______________________________________________
>> ARIN-Discuss
>> You are receiving this message because you are subscribed to
>> the ARIN Discussion Mailing List (ARIN-discuss at arin.net).
>> Unsubscribe or manage your mailing list subscription at:
>> http://lists.arin.net/mailman/listinfo/arin-discuss
>> Please contact info at arin.net if you experience any issues.
> _______________________________________________
> ARIN-Discuss
> You are receiving this message because you are subscribed to
> the ARIN Discussion Mailing List (ARIN-discuss at arin.net).
> Unsubscribe or manage your mailing list subscription at:
> http://lists.arin.net/mailman/listinfo/arin-discuss
> Please contact info at arin.net if you experience any issues.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.arin.net/pipermail/arin-discuss/attachments/20100913/706ea89d/attachment.html>


More information about the ARIN-discuss mailing list