[rbldnsd] generic dataset still supported?

[Michael Tokarev]

nathan r. hruby wrote:
> Hi Folks,
> 
> Is the "generic" dataset type still supported and planned to be supported
> in rbldnsd?

Well, it's an integral part of the whole thing.  So.. yes.  I hope ;)

> I'm wondering because I have a possible "interesting" application that
> requires about 80k A records in a zone and will need to be regenerated
> every hour, using the generic format.  I have doubt about bind's ability
> to do this well, but I know that rbldnsd can.

The generic dataset type is rather simplistic, and isn't optimized very
well.  It's aimed as an addition for main datasets like ip4set et al,
just to add "a little" of DNS metadata, say, a few records like MX for
the domain, maybe an A record for a web page, maybe some descriptional
TXT record and so on.  I mean, I can't guarantee it will work well with
large amount of records, like other "main" datasets handle.  It isn't
"bad", but I just never thought about using it in bulk.

On the other hand, 80k isn't that much really.  On modern hardware, I'd
say any nameserver is capable to handle that stuff pretty easy.  Why do
you think BIND will be unable to handle it?  Just try it and see?

And also, if you only want A records, try using dnset instead. It should
work a bit better (in terms of memory and speed) than generic dataset,
and - probably, really depending on your usage - should just work for
you.  But again, for 80k entries, there should be no big difference.

/mjt