From brett at the-watsons.org Wed Sep 1 01:59:36 2010 From: brett at the-watsons.org (Brett Watson) Date: Tue, 31 Aug 2010 21:59:36 -0400 Subject: [Bind10-uides] Apologies Message-ID: <2BD26B46-F4E5-454D-B15C-653C94226AB5@the-watsons.org> I've been traveling for about 1.5 weeks. Trying to catch up on threads, apologies for no response. -b From scharf at isc.org Wed Sep 1 06:09:23 2010 From: scharf at isc.org (Jerry Scharf) Date: Tue, 31 Aug 2010 23:09:23 -0700 Subject: [Bind10-uides] proof of cluelessness and a general question Message-ID: <4C7DEE13.5060504@isc.org> Hi folks, I presented some of the basic ideas and both Shane, the person who I work for, and the team seem in general to like it. One person said that the macro system that DHCP uses is worth looking at. "Once we explain to them what to add and where to add it, they can get their jobs done quite well." Damning with faint praise. I am afraid I was not too kind in response. One of the people who had the highest risk of having a cow was quite supportive in principal and wanted to start wordsmithing the terms I had chosen. He now wants the management data model/store to be logically part of the server. I don't think he understands the full implications of this (botnet C&C, here we come.) At least this time I get to spec it, create the road map and then run away. :) So I read the BIND 10 specs and I hear people like Ben talk about RESTful interfaces. The way I read it, if I want to have a login session or worse open a set of commands to be treated as a transaction, I need to defeat the concept of RESTfullness somewhere. Am I missing something or is this one of the less useful design concepts for a management tool? A second and more serious question. How do things like virtualization and cloud computing change the way DNS is used and the object structures that the management tool needs to cope with? I have a suspicion that this is one of those things that either we put some hard thought into or it will bite us when we roll it out. jerry