RWHOIS

[Shane Kerr]

On 2001-04-27 14:51:13 +0000, Mark Kosters wrote:
> On Fri, Apr 27, 2001 at 02:26:34PM -0400, Richard Jimmerson wrote:
> > ARIN has determined taking over the future support and development
> > of RWHOIS may require a complete rewrite of the software. There has
> > also been some discussion on this list about the possibility of a
> > third reassignment option. Ginny Listman will be moving this
> > discussion over to the dbwg at arin.net mailing list.
> 
> I don't think it needs a complete rewrite. I do however think that the
> db stuff on the backside needs to be replaced with mysql, oracle, or
> something else. People can use different tools to update and display
> info that resides on their database other than using rwhois or
> cumbersome text editing that is prone to error.

Having looked through the code, I can say that Mark is right - there's
not a definite need for a complete rewrite.  

In addition to changing the backside, the front-end will probably need
to be revisited as well.  The current fork & exec model of the ARIN
Whois server works well because the client interface is extremely
simple, and the database is very tightly optimized for the specific
query patterns.  But the RWhois service does not share these properties.
I suggest that you'd probably want (in order of effeciency):

1. stacks of computers for your fork/exec server
2. a multithreaded process
3. an event-driven single process

I do think it is a bad idea to maintain two seperate databases that do
effectively the same thing.  For instance, the Whois and RWhois servers
maintain seperate limits on the number of results returned - but in
different locations and with different semantics.  No need to discuss
the obvious disadvantages of this duplication, I think.

Shane