misunderstanding scale (was: Ipv4 end, its fake.)

Mark Tinka mark.tinka at seacom.mu
Sun Mar 23 19:02:05 UTC 2014


On Sunday, March 23, 2014 08:35:48 PM Saku Ytti wrote:

> Or IT isn't buying the 'renumbering is easy' argument,
> for any non-trivial size company even figuring how where
> exactly can be IP addresses punched out statically would
> be expensive and long process.
> If you are pushing for customer to use your PA in their
> LAN, I'm guessing net-result is you should never reclaim
> those addresses after customer leaves, since chances
> are, some customers won't renumber, but will 1:1 NAT
> your PA to new operator PA, and your next customer with
> this block will complain about reachability problems to
> this other customer.

In all fairness, I'm not so sure, as operators, that we want 
to push our PA space as assignments to customers in IPv6-
land.

Yes, it makes sense, but then again, it's not hard for 
enterprises to obtain PI space from $favorite_registry. Yes, 
that will pollute the routing table and potentially mean 
your customer can run away from you at any time. But IPv6 is 
so vast, and as you rightly point out, Saku, it might be 
unreasonable for us to expect the enterprise to renumber 
when they churn and take their business elsewhere. It, 
physically, is a lot of work.

So while I have lots of /56's and /48's to assign to 
customers from my /32, I'm not sure I want to actively 
encourage it, unless as a last resort.

Of course, assigning this to broadband users makes more 
sense, as use is generally temporary and well controlled.

Mark.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part.
URL: <http://mailman.nanog.org/pipermail/nanog/attachments/20140323/4922e815/attachment.bin>


More information about the NANOG mailing list