RWHOIS

[Shane Kerr]

On 2001-04-27 15:11:04 +0000, Mark Kosters wrote:
> On Fri, Apr 27, 2001 at 04:55:45PM +0200, Shane Kerr wrote:
> > The new RIPE Whois server, in production since April 23, is Much
> > Improved(tm).  Currently it only supports referrals for several types of
> > records - network allocations (inetnum) are not among them.  It would
> > probably be a lot less work to add this capability than to maintain two
> > seperate database interfaces.  Perhaps ARIN should pursue this as a goal
> > instead of taking on maintaining RWhois?  After all, the RIPE Whois
> > server is very actively maintained.  ;)
> 
> I'm glad to hear that RIPE has improved their database.

Believe me, so am I!

> Personally, I'd like to see the data distributed and not reliant on a
> box (or boxes) that have to chase referrals for the distributed info
> to send back to the client.  We do this at the Verisign registrar for
> the gtld whois stuff and it is non-trivial.  Resources to do this at
> fairly high volume rates is pretty expensive. 

True...

> Work is on-going to replace this creaky system. I hate to see ARIN
> follow this model and caught a couple of years from now with volumes
> that they may not want to be able to sustain*.

It is still possible to write a client that follows referrals under the
RIPE database.  Sending a "-R" before the query tells the server not to
follow referrals.  In this manner, the client can see the record that
documents the referral and follow it without the server's help.

The RIPE server already supports an SQL backend (MySQL), is
multithreaded, has a sophisticated dynamic rate limiting functionality,
and works with plain-old Whois (POW?) clients.  In all likelyhood, it is
going to be used by at least 3 of the 5 pending RIR's (RIPE NCC, APNIC,
and LACNIC - not too sure about AFRINIC, but I suspect they'll use it as
well).

It doesn't have some of the cooler features of RWhois, like querying the
schema, and it has a lot of design cruft from olden days (the maintainer
mechanism springs immediately to mind).  

I guess in my mind from the user point of view, there's no difference
between an RWhois server that follows referrals and a Whois server that
follows referrals.  The only real difference would be if the server was
like the Internic Whois server and didn't follow referrals at all, in
which case the client would have to do the work.  As it is, either
RWhois or Whois can function in that manner as well.

But from the ISP point of view, they may prefer to *run* the RWhois
server, because it requires less investment in installation.  There's
nothing to prevent a Whois referral from ARIN going to an RWhois server,
of course....

Shane