The role of reverse zone files

[Steve Sandau]

David Botham wrote:
> bind-users-bounce at isc.org wrote on 11/08/2004 10:50:54 PM:
> 
>>Ok some of the dumber fog is starting to lift
>>Clearly, knowing an address does not tell me how to find a zone
>>file, forward or reverse. There has to be a reverse chain to go 
>>with the forward chain for external queries. 
>>
>>I still don't quite get _how_, given an arbitrary address, the
>>correct inverse zone file can be found to consult for the hostname.
> 
> 
> 1.  Start with an ip address, say, 192.168.2.5
> 2.  Create a domain name with by:
>         a.  Reversing the order of the octets to yield 5.2.168.192
>         b.  Append .in-addr.arpa to the end of it to get: 
> 5.2.168.192.in-addr.arap.
> 3.  Start at the root asking this question:
>         5.2.168.192.in-addr.arap PTR
> 4.  Get a referral to the ISP who owns the 192.168.2.0 network:
>         2.168.192.in-addr.arap NS ns1.someisp.com
>         2.168.192.in-addr.arap NS ns2.someisp.com
> 5.  Ask ns1.someisp.com the same question:
>         5.2.168.192.in-addr.arap PTR
> 6.  Get the answer from their zone file:
>         5.2.168.192.in-addr.arap PTR somehost.someispornot.com
> 
> 
> Steps 1 and 2 are automatically performed by resolvers.
> 
> The remainder of the process is the same for forward zones as it is for 
> reverse zones.  In fact there is little difference between them from that 
> perspective, except for the type of information typically stored in the 
> zone.
> 
> 
>>The key has to be in the chain of in-arpa domains, but I don't
>>see the connection between my in-arpa zones and my provider's
>>DNS. Wait a minute, is this what's meant by "reverse delegation"?
> 
> 
> Reverse delegation is a term some people use to refer to the delegation of 
> reverse zones.  There is no difference in the way reverse and forward 
> zones are delegated, with the exception of RFC2317.  Even RFC2317 performs 
> the actual delegation in the same manner.
> 
> 
>>Essentially, my ISP saying to the world "these numbers are resolved
>>by a nameserver at address w.x.y.z?
> 
> 
> Numbers are not resolved.  Domain names are resolved.  The domain names 
> for "reverse" zones *look* a lot like numbers, but, they are simply labels 
> in the domain name.  See steps 1 and 2 from above for the process required 
> to derive the correct domain name to query given a particular starting IP.
> 
> 
>>This is a tougher homework assignment than I bargained for.....
>>If the logic is not in Hunt's book please warn me and I'll spring
>>for the other one.
> 
> 
> Get a copy of DNS & BIND 4th Edition.
> 
> 
> 
>>I don't think I'm dealing with a /24 network, which would imply
>>that I have 8 bits of address space. The netmask given is 
>>255.255.255.248, which appears to give me three bits or 7 
>>distinct addesses. Five are said to be "assignable", one is
>>occupied by the router. Not sure what became of the last one.
> 
> 
> If you are wondering how you would control the in-addr.arpa name space 
> associated with the block of IP addresses you have, you should consult 
> RFC2317.
> 
> 
> hth,
> 
> 
> Dave...
> 

All said better than I could say it. :)

Additionally, somewhat off-topic, the .248 netmask gives you 256-248 
addresses, or 8 total. The lowest address is the network address, and 
the highest is the broadcast address. Both are unusable for hosts. This 
leaves 6 addresses for hosts, one of which is used by the router.

Steve Sandau