Wildcards in reverse DNS

Edward Lewis Ed.Lewis at neustar.biz
Mon Jan 8 13:55:10 UTC 2007

At 8:47 +1100 1/6/07, Mark Andrews wrote:

>	How could it be the fault of the protocol designer when the
>	properties of the network have changed underneath the
>	protocol designer.  Most of the protocols were written when
>	IP addresses in IP header didn't change between source and
>	destination.  The packets had enough unique information to
>	get the responses back to the originator.

The "fault" lies in violating layering.  But this isn't "fault" as in 
"you are guilty of a crime against humanity" but a "fault" like 
failing to anticipate the future correctly.

I once marvelled at a very simplistic Mac application that was 
written in 1985 or 1986 called "RoadAtlas".  If calculated routes and 
distances along US interstate highways and other major roads between 
US cities.  It ran on every Mac platform up through Mac 9, and then 
on pre-Intel Macs using version 9 emulation in MacOS X.  On the other 
hand, there were many "professionally" written applications that 
would run on Mac 5.4 but not a newer version.

Sometimes something is so perfectly written, or solves a problem so 
tightly constrained, it can last into the future.

>	No.  It mean that if you have legacy code it needs a minor
>	re-write.  Something that most programers could do in a
>	afternoon.

Assuming that you are a programmer.  Not many are.

>	Both IPv4 and IPv6 have routing issues.  They are roughly
>	the same.  However IPv6 was designed to ease the problems
>	of renumbering which should, in theory, relieve some of the
>	routing issues.

Both have issues, but the problems in IPv4 are already known and 
being dealt with.  IPv6 issues are new and there is fear about them. 
I'm referring to the sentiment of last April when the North American 
RIR supported allotting "provider independent" address space over the 
objections of both ISPs and IETF protocol designers.

They issues are not roughly the same.  IPv4 has already burned the 
bridge and because of the smaller size of the address space, ISPs are 
not afraid.  IPv6 hasn't ventured far into the swamp.  If you want to 
see what makes ISP afraid and why I think they have reason to fear 
this, there are some presentations given by a Jason Schiller of MCI 
at ARIN and NANOG meetings.

>	One of the biggest problem is that people try to apply IPv4
>	solutions to IPv6 rather than take advantage of what IPv6
>	offers.  IPv6 addresses lots of problems identified with
>	IPv4, not just the number of addresses.

I don't think that applying IPv4 techniques to IPv6 is a "problem" 
but a design constraint.  People who get used to one way of doing 
business don't what to have to relearn new tricks for an upgrade. 
Not every one using the Internet is a network engineer.

>	NAT is a IPv4 solution to a IPv4 problem.  IPv6 eliminates
>	the need to do NAT.

Probably so, but there has been resistance to DHCPv6, saying that it 
is a vestige of IPv4 not needed in an autoconfiguring world.  (That 
is something network managers beg to differ on, they like the DHCP 
style control they have now.)
Edward Lewis                                                +1-571-434-5468

Dessert - aka Service Pack 1 for lunch.

More information about the bind-users mailing list