Re: IPv6 Prefix announcing

2011-04-26 Thread Justin M. Streiner 

On Tue, 26 Apr 2011, Nick Olsen wrote:


I've always been under the impression its best practice to only announce
prefixes of a /24 and above when it comes to IPv4 and BGP.
I was wondering if something similar had been agreed upon regarding IPv6.
And if That's the case, What's the magic number? /32? /48? /64?


You're likely to get different answers to this, but the 'magic number' 
appears to be /48.  Looking in the v6 BGP table, you will likely find 
smaller prefixes than that, but a number of the major carriers seem to be 
settling on /48 as the smallest prefix they will accept.  /48 is also the 
smallest block most of the RIRs will assign to end-users.


jms

RE: IPv6 Prefix announcing

2011-04-26 Thread Kate Gerry 
Funny enough, some carriers actually require the 'smallest' as being /32... :(


-Original Message-
From: Justin M. Streiner [mailto:strei...@cluebyfour.org] 
Sent: Tuesday, April 26, 2011 9:34 AM
To: nanog@nanog.org
Subject: Re: IPv6 Prefix announcing

On Tue, 26 Apr 2011, Nick Olsen wrote:

> I've always been under the impression its best practice to only 
> announce prefixes of a /24 and above when it comes to IPv4 and BGP.
> I was wondering if something similar had been agreed upon regarding IPv6.
> And if That's the case, What's the magic number? /32? /48? /64?

You're likely to get different answers to this, but the 'magic number' 
appears to be /48.  Looking in the v6 BGP table, you will likely find smaller 
prefixes than that, but a number of the major carriers seem to be settling on 
/48 as the smallest prefix they will accept.  /48 is also the smallest block 
most of the RIRs will assign to end-users.

jms

Re: IPv6 Prefix announcing

2011-04-26 Thread Patrick W. Gilmore 
On Apr 26, 2011, at 12:39 PM, Kate Gerry wrote:

> Funny enough, some carriers actually require the 'smallest' as being /32... :(

Vote with your wallet.

Some carriers would prefer if only transit free networks were allowed to 
originate routes.  Doesn't mean you should follow their lead.

-- 
TTFN,
patrick


> -Original Message-
> From: Justin M. Streiner [mailto:strei...@cluebyfour.org] 
> Sent: Tuesday, April 26, 2011 9:34 AM
> To: nanog@nanog.org
> Subject: Re: IPv6 Prefix announcing
> 
> On Tue, 26 Apr 2011, Nick Olsen wrote:
> 
>> I've always been under the impression its best practice to only 
>> announce prefixes of a /24 and above when it comes to IPv4 and BGP.
>> I was wondering if something similar had been agreed upon regarding IPv6.
>> And if That's the case, What's the magic number? /32? /48? /64?
> 
> You're likely to get different answers to this, but the 'magic number' 
> appears to be /48.  Looking in the v6 BGP table, you will likely find smaller 
> prefixes than that, but a number of the major carriers seem to be settling on 
> /48 as the smallest prefix they will accept.  /48 is also the smallest block 
> most of the RIRs will assign to end-users.
> 
> jms
> 
>

RE: IPv6 Prefix announcing

2011-04-26 Thread George Bonser 



> From: Kate Gerry 
> Sent: Tuesday, April 26, 2011 9:39 AM
> To: 'Justin M. Streiner'; nanog@nanog.org
> Subject: RE: IPv6 Prefix announcing
> 
> Funny enough, some carriers actually require the 'smallest' as being
> /32... :(
> 

That might be true in PA space, but PI space is issued down to /48.  I
am not aware of anyone who filters smaller than a /32 in PI space though
that doesn't mean it doesn't happen.  The largest holdout was Verizon
but my understanding is they now accept a /48 in PI space.

So: 

A /32 is the smallest prefix issued in PA and some networks will not
accept a prefix smaller than /32 from PA address space.
A /48 is the smallest prefix issued in PI and some networks will not
accept a prefix smaller than /48 from PI address space.

In other words, if you are going to attempt to multihome a /48
allocation from your provider's aggregate, you are better off getting
your own provider independent block.

Re: IPv6 Prefix announcing

2011-04-26 Thread William Herrin 
On Tue, Apr 26, 2011 at 12:30 PM, Nick Olsen  wrote:
> Greetings NANOG,
> I've always been under the impression its best practice to only announce
> prefixes of a /24 and above when it comes to IPv4 and BGP.
> I was wondering if something similar had been agreed upon regarding IPv6.
> And if That's the case, What's the magic number? /32? /48? /64?

Hi Nick,

At this point, you can depend on being able to announce a /32 from any
block and a /48 from an RIR block designated for end-user assignments.
Many carriers have more permissive policies but all of any consequence
now allow at least that.

Regards,
Bill Herrin


-- 
William D. Herrin  her...@dirtside.com  b...@herrin.us
3005 Crane Dr. .. Web: 
Falls Church, VA 22042-3004

Re: IPv6 Prefix announcing

2011-04-26 Thread Owen DeLong 
I know that used to be true, but, to the best of my knowledge, everyone is now 
accepting
down to /48s in provider independent ranges. Some still require /32 or shorter 
in the provider aggregate ranges.

Owen


Sent from my iPad

On Apr 26, 2011, at 10:39 AM, Kate Gerry  wrote:

> Funny enough, some carriers actually require the 'smallest' as being /32... :(
> 
> 
> -Original Message-
> From: Justin M. Streiner [mailto:strei...@cluebyfour.org] 
> Sent: Tuesday, April 26, 2011 9:34 AM
> To: nanog@nanog.org
> Subject: Re: IPv6 Prefix announcing
> 
> On Tue, 26 Apr 2011, Nick Olsen wrote:
> 
>> I've always been under the impression its best practice to only 
>> announce prefixes of a /24 and above when it comes to IPv4 and BGP.
>> I was wondering if something similar had been agreed upon regarding IPv6.
>> And if That's the case, What's the magic number? /32? /48? /64?
> 
> You're likely to get different answers to this, but the 'magic number' 
> appears to be /48.  Looking in the v6 BGP table, you will likely find smaller 
> prefixes than that, but a number of the major carriers seem to be settling on 
> /48 as the smallest prefix they will accept.  /48 is also the smallest block 
> most of the RIRs will assign to end-users.
> 
> jms
>

Re: IPv6 Prefix announcing

2011-04-26 Thread Seth Mattinen 
On 4/26/2011 09:39, Kate Gerry wrote:
> Funny enough, some carriers actually require the 'smallest' as being /32... :(
> 

This is becoming the exception now, not the rule.

Last year I was fighting with Verizon about their refusal to carry /48s.
That, together with the impasse of figuring out how to put dual stack
IPv6 on an Ethernet port (it was delivered as IPv4 only multiple times),
I never accepted it and went with a competitor who got it right the
first time. However, I've had several sources tell me Verizon has since
backpedaled and now accepts /48s.

~Seth

RE: IPv6 Prefix announcing

2011-04-26 Thread Michael K. Smith - Adhost 
> -Original Message-
> From: Seth Mattinen [mailto:se...@rollernet.us]
> Sent: Tuesday, April 26, 2011 12:52 PM
> To: nanog@nanog.org
> Subject: Re: IPv6 Prefix announcing
> 
> On 4/26/2011 09:39, Kate Gerry wrote:
> > Funny enough, some carriers actually require the 'smallest' as being /32... 
> > :(
> >
> 
> This is becoming the exception now, not the rule.
> 
> Last year I was fighting with Verizon about their refusal to carry /48s.
> That, together with the impasse of figuring out how to put dual stack
> IPv6 on an Ethernet port (it was delivered as IPv4 only multiple times),
> I never accepted it and went with a competitor who got it right the
> first time. However, I've had several sources tell me Verizon has since
> backpedaled and now accepts /48s.
> 
> ~Seth

*> 2001:67C:120::/482001:504:16::1B1B   150  0 6939 701 12702 43751 
6716 i

Mike