how is a named resolved?

Kevin Darcy kcd at daimlerchrysler.com
Thu Apr 6 03:02:41 UTC 2000


The flow would normally be Figure 2, which shows normal iterative resolution.
What is probably confusing you is that sometimes people set up their
nameservers to "forward" to other nameservers, which means they act as clients
and expect the forwarder to resolve the query for them. If B (a .com
server) and C (a server for bbb.com) were setup in a forwarding hierarchy
pointing towards the root, then Figure 1 would be correct. But such forwarding
hierarchies are not feasible on the Internet, since the root servers and the
.com servers do not perform recursion. Forwarding hierarchies are therefore
normally only found on intranets, and usually because the internal clients
need to resolve Internet names, but only certain nodes on the network (e.g.
firewalls or nameservers in the DMZ) have the necessary connectivity to
resolve those names.


- Kevin

flybird wrote:

> Hi,all
> I have a question about how named is resolved.
> For example, there is a DNS server "A" which manage a domain and is a cache
> server for root domain. So while it get a query for aaa.bbb.com, it will
> query root DNS "B". And suppose DNS for com is "C", DNS for bbb.com is "D".
> And than will DNS "B" query the "C" for aaa.bbb.com or it just reply to "A"
> that DNS for com is "C" and then "A" query "C" for aaa.bbb.com ?
> There are tow figure below to help me express my question:
>
>  1    2    3
> ---> ---> --->
> A    B    C    D
> <--- <--- <---
>  6     5    4
>     Figure 1
>
> OR
>  1
> --->
> A    B
> <---
>  2
>
>  3
> --->
> A    C
> <---
>  4
>
>  5
> --->
> A    D
> <---
>  6
> Figure 2
>
> In my impression, the flow of DNS query is Figure 1. But After i saw the
> discuss "Private Public DNS question" in this group, it seems the flaw is
> Figure 2. Now i don't know which one is true.
> Thanx for any help,
>
> Pan Tao






More information about the bind-users mailing list