[bind10-dev] BIND 9 compatibility, was BIND 10 face to face meeting Y2-2 agenda

[Shane Kerr]

Fujiwara-san,

On Fri, 2010-08-27 at 19:39 +0900, Kazunori Fujiwara wrote:
>   - view/shared socket issue
> 
>       A socket may be shared with authoritative service and recursive
>       service on BIND 9.
> 
>       # I think it is not a good model.
> 
>       To support socket sharing, Auth server model need to be changed.
> 
>       socket-->[Dispatcher/Distributer]-->[ACL]-->[Auth]  per zone?
>                per view, per zone      -->[ACL]-->[Auth]
>                                        -->[ACL]-->[Recursion]
> 
>       I think we need to consider auth and recurive server's socket model.
>       I prefer separating model, but it is not BIND 9 compatible.

We prefer separate tasks, like you describe, but need to see if it is
efficient to do that. This is the main point of Stephen's recent
research, which he will be presenting & discussing with us on Tuesday
morning. It looks like we may be able to have separate tasks.

This brings a question as to what we mean by BIND 9 compatibility in
general.

I think that we do NOT need to make BIND 10 look like BIND 9 in terms of
the processes that are running. My idea is that we need to make sure
that people running in BIND 9 compatibility mode can use the same
configuration files and commands that they use now. So, named.conf
loads, you start things up using "named -4 -p 5353" or whatever you use,
you can continue to use rndc, and you can send SIGHUP to the named
process to reload things.

We don't need to worry that we're running processes in the background,
creating hidden inter-process communication and other data structures,
or things like that. These are things that administrators only need to
worry about if they plan on taking advantage of new BIND 10
functionality.

If nothing else, BIND 10 will be different from BIND 9 because it will
be faster. ;)

> # I think there are another many issues.

I agree! :)

--
Shane