[161053] in North American Network Operators' Group
Re: can you share ipv6 addressallo cation
daemon@ATHENA.MIT.EDU (Deric Kwok)
Sun Feb 24 12:01:12 2013
In-Reply-To: <56982C89-3F69-41BD-9097-C1E268519D3D@delong.com>
Date: Sun, 24 Feb 2013 12:00:58 -0500
From: Deric Kwok <deric.kwok2000@gmail.com>
To: Owen DeLong <owen@delong.com>
Cc: nanog list <nanog@nanog.org>
Errors-To: nanog-bounces+nanog.discuss=bloom-picayune.mit.edu@nanog.org
Hi Owen and all
Thank you so much for all info. I do have question about /126 or /64
as link to link
After getting this
http://www.clarksys.com/blog/2010/08/18/howto-subnet-ipv6-for-network-links=
/
Can I know what do you think?
Can we say that to use /64 to replace /126 for network link?
and what do you think the problem to use /64?
The website said:
If you refer back to the presentation I mentioned earlier there=92s
notes about the potential dangers of /64s on network links and why
people intuitively want to subnet to a /127 or a /126. We ended up
splitting the difference and actually subnetting the /64 for the
network link to a /126.
IPv6 is a very large pool of IP space =96 to paraphrase my favorite
quote so far =93IPv6 has 340 undecillion unique addresses (that=92s a
39-digit number). If IPv4 is a golf ball IPv6 is the sun.=94 Trust me,
don=92t try to over think this. Just follow what all the RFCs say and
use /64s for your network links.
If you want to read more I found the following links to be very
helpful in understanding how to properly subnet IPv6:
Thank you so much
On Wed, Feb 20, 2013 at 9:00 PM, Owen DeLong <owen@delong.com> wrote:
> First, if you are starting from a /32 and deciding how to carve it up fro=
m there, you are already approaching the problem backwards.
>
> The correct approach (general broad strokes) is to:
>
> 1. Identify your subnetting needs.
> A. Infrastructure addressing
> B. Internal IT needs within the company
> C. Customer network needs (usually best to count the=
Infrastructure and Internal IT as n*customers at this point when
> rolling this all up into a total number of subnet=
s needed).
> D. Decide on a customer end-site subnet size (unless=
this is an exceptional case, /48 is a good number to use)
>
> 2. Identify the natural aggregation points in your network.
>
> 3. Identify the number of /48s (or whatever other size you d=
ecided in D) needed
> in your largest aggregation site. (This should be the sum=
of all subordinate
> end-user networks as well as any infrastructure networks,=
etc.
>
> Round that up to a nibble boundary ensuring at least a 25=
% free space.
>
> 4. Identify the total number of aggregation points at the hi=
erarchy level identified in (3) above.
>
> 5. Round that up to a nibble boundary as well.
>
> 6. Make a request for the prefix size determined by taking t=
he number in 1D (/48) and
> subtracting the number of bits identified in (3) and (5).=
e.g. your largest aggregation
> point serves 50,000 customer end sites and you have 196 s=
uch aggregation points.
> Each customer end-site is to receive a /48.
>
> 50,000 customer end-sites is 16-bits. To get a 25% min fr=
ee, we must round up to 20.
> This count includes 2 customer end-sites to support ISP i=
nfrastructure and internal IT
> needs, respectively.
>
> 196 aggregation points is 8-bits. To get a 25% min free, =
we must round up to 12.
>
> 48-20=3D28-12=3D16 -- This network should request a /16 f=
rom their RIR.
>
> Notes:
>
> This is a severe oversimplification. Obviously more details will be requi=
red and the process must be adapted to each individual ISP's network topolo=
gy and other considerations.
>
> Your first several iterations of addressing plan will be wrong. Accept it=
, deploy it, and expect to redo it a few times before you're completely hap=
py with it.
>
> Plan big, deploy small the first few times so that you can learn lessons =
about the big plan while the deployments are still small.
>
> Owen
>
> On Feb 20, 2013, at 14:44 , Deric Kwok <deric.kwok2000@gmail.com> wrote:
>
>> Hi all
>>
>> I am searching information about ipv6 addressallocation for /32
>>
>> Any experience and advice can be shared
>>
>> eg: loopback. peer to peer,
>>
>> Thank you so much
>
