I have asked that a discussion of rwhois and ARIN's current and future support of rwhois be discussed at the next ARIN members meeting on October 16, 1998. This item has been added to the agenda for that meeting. I will be glad to present any questions/points/items about the future of rwhois, please email me privately with this information. Thanks, Ron |||||||||||||||||||||||||||||||||||||||||||||||||||\ Ron Muir IP Administration BellSouth.net .net 770-522-6363 ron.muir@bellsouth.net - ----------

From: Leo Bicknell <bicknell@dimension.net> To: nanog@merit.edu Subject: rwhois Date: Wednesday, September 16, 1998 8:16 PM

I know quite a few of you were interested in my rwhois script, and I have exchanged a lot of e-mail with some of you individually. I think it's time I gave a bit of a summary.

First off, it's definately the right idea. Properly executed this could provide a seamless, distributed method for looking up whois data from all the registries. If we could get past the problems below it would be a dream for network operators.

Of course, there are problems. I'll enumerate these in the order of importance in my mind, others may disagree.

1) Data is missing/inaccurate. root.rwhois.net only reliably provides Internic data, and that data is older than whois data. ARIN does not yet participate in rwhois, and as I understand it RIPE and APNIC have not yet tied in. There is no .US data in rwhois either.

Without the right data in here it's useless, regarless of all other problems.

2) The server is not completely developed. There are features in the server that are not fully implemented. Nuff said.

3) There is no distributed scheme set up. The Internic runs one "root" rwhois server. This is no better than whois. There should be 5-10 distributed across the network to provide redundancy.

4) The rwhois protocol needs a caching server. It should be possible for any ISP to set up a caching/mirroring rwhois server with a relatively few steps so they can have their own local server to query. This allows scripts to pound on the data without affecting people outside of that ISP.

So, the net result is I don't see how an ISP could use rwhois today. With one server at the same location as the whois server it's no more "reliable", and the data is definately behind. Plus they actually have tigher limits (eg number of results returned) on the rwhois server than the whois server.

I really think some people should get behind rwhois and push it forward, as it is a reasonable solution, and is probably 90% of the way developed. If only it could be deployed and supported.

-- Leo Bicknell - bicknell@dimension.net Network Engineer (CCIE #3440) - Dimension Enterprises 1-703-709-7500, fax, 1-703-709-7699