Being the person responsible for the agenda for the September meeting, I hereby ask for agenda items, and general input. As we all know, the DNS wg has been not as focused as would be expected by a wg at RIPE. The new chairs want to change that. As you can see on http://www.ripe.net/ripe/wg/dns/index.html, the charter is extremely limited, and on top of that wrong:
The Domain Name System working group discusses current BIND versions. It is also concerned with potential pollution of the DNS and with domain name related issues.
The DNS working group has the last couple of meetings had two goals: - Discussion forum for "current DNS operational issues" - Presentations on status for various project/initiatives elsewhere, such as the DNS related topics in the IETF If it is the case this wg feel there is space for both of the above (something I personally hope), the next step is to focus each one of them a bit more. Especially the first one. The meta-discussion I would like to see in Amsterdam, with a conclusion at the end, is what current DNS operational issues include. The answer should be in the form of a list of potential deliverables, results from discussions on this mailing list and at the meetings. I want a designated individual responsible for each item, and further, a small group of dedicated people which has as main task of moving the discussion forward. The topics are to be synchronised with what happens in DNR Forum, and because of this, you see DN* on the agenda for the next RIPE meeting. The division of labour between the two groups (if we will still have two groups in the future) must be more clear. CENTR today exists as a separate organisation just like anyone else, so maybe the DNR activities should be rolled into the DNS wg (i.e. the two groups should be merged). One of the topics could of course be report from CENTR what is happening, just like other bodies can do it. Proposed agenda for the meeting is because of this: Part 1: Initial discussion about the charter of the DNS wg - What are we doing here? - If we should talk about "operational issues/experiences with DNS", what can be the deliverables? + Topic: Deliverable: Topic-master: Group-members: - What bodies are interesting to have reports by, part from the Topic-masters? Part 2: Reports from DNS-related organisations (2 minutes each): [This is only a list I create from the top of my head...no one has confirmed or even asked to be on the agenda] - IETF: DNSOP - IETF: DNS - Root-server operators - CENTR technical committee - ccTLD registries - Registrars Conclusion: - What is the new wording of the charter? Comments are appreciated. This is _your_ wg, and we are to work together to reach some goal which is of benefit for all of us. paf
Patrick, I would like to have approx 30 minutes to present the pages at dnsmon.ripe.net which will by then be in beta test moving to production. Daniel
On torsdag, jul 17, 2003, at 15:01 Europe/Stockholm, Daniel Karrenberg wrote:
I would like to have approx 30 minutes to present the pages at dnsmon.ripe.net which will by then be in beta test moving to production.
Ack. Do you have any proposal what "deliverables" this can lead to? What can the conclusions be? Recommendations on how/where to build DNS servers, or a document on "trying to measure quality in these ways does NOT reflect real world"? I.e. can we turn this recurrent presentation into some more general task for the wg? paf
On torsdag, jul 17, 2003, at 21:02 Europe/Stockholm, Patrik Fältström wrote:
On torsdag, jul 17, 2003, at 15:01 Europe/Stockholm, Daniel Karrenberg wrote:
I would like to have approx 30 minutes to present the pages at dnsmon.ripe.net which will by then be in beta test moving to production.
Ack.
Do you have any proposal what "deliverables" this can lead to? What can the conclusions be? Recommendations on how/where to build DNS servers, or a document on "trying to measure quality in these ways does NOT reflect real world"?
I.e. can we turn this recurrent presentation into some more general task for the wg?
He he he...at last I managed to stir up some dust. Let me clarify something in the text above. Just like Daniel has pointed out several times in all of his presentations, what Daniel tests is still only the quality of each one of the servers as viewed from the measuring points. It is extremely unclear what conclusions can be made (if any) about the DNS *SERVICE* which the servers deliver together to a random user on the Internet. So, a proposed task might be for this wg is to look at the tool created, and many many other reports (including maybe the report Richard delivered at the last EOF about full service resolver behaviour), and deliver a paper which says _how_ to evaluate where DNS servers should be located. People argue all the time what is "the best" location of servers. Could this not be a good discussion? (It seems to be a good discussion point... ;-) Any takers on being token holder for this "task"? paf
On 17.07 21:02, Patrik Fältström wrote:
On torsdag, jul 17, 2003, at 15:01 Europe/Stockholm, Daniel Karrenberg wrote:
I would like to have approx 30 minutes to present the pages at dnsmon.ripe.net which will by then be in beta test moving to production.
Ack.
Do you have any proposal what "deliverables" this can lead to? What can the conclusions be? Recommendations on how/where to build DNS servers, or a document on "trying to measure quality in these ways does NOT reflect real world"?
I.e. can we turn this recurrent presentation into some more general task for the wg?
1) Recommendations for measures DNS serv*ice* quality 2) Recommendations how to summarise the results for consumption by the non-technical public.
On fredag, jul 18, 2003, at 13:43 Europe/Stockholm, Daniel Karrenberg wrote:
1) Recommendations for measures DNS serv*ice* quality
2) Recommendations how to summarise the results for consumption by the non-technical public.
Thanks! paf
At 9:59 +0200 7/10/03, Patrik Fältström wrote:
Being the person responsible for the agenda for the September meeting, I hereby ask for agenda items, and general input.
Please catagorize this under "general input." ;) At the last meeting, DNS lameness was a topic. Unfortunately I have not dedicated enough time on my end to make significant progress on my part, but I am wondering if there has been some in other areas. (It's still 6 weeks off, so maybe I can do something.) Recommending an approach to stamping out lame delegations is a high goal, and I think finding a "grand unified approach" that satisfies all is too big of a first step. Two smaller steps that I think are attainable are: 1) Defining what is meant by the word "lame" in the operational context - given that "lame delegations" are used in IETF RFCs to describe a state that is a subset of the operational problem. 2) Defining a way to measure the extent of the problem. I.e., Given a zone: How many delegations, NS RR's, and servers are present? What % of zones cannot be contacted at a given a moment? What % of servers ... the same. What % of NS RR's are a problem... Note that #1 really is needed before #2 - because, what is a "problem?" ;) One comment on this is that I think we can answer steps 1 and 2 given different registry policies. E.g., some registries are happy simply if you can get an answer from a delegation. On the other hand, some registries won't be happy unless the responsible person answer mail and that the MX RR does not refer to an open relay. I'm suggesting that this might be a seed of an agenda item (but I can't now promise to deliver an agenda item). PS - Ordinarily, being policy neutral is a goal of any engineering effort. Perhaps though, it would be good to go the other way - try and describe various policies so that we each have a broader view. Shrug. -- -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- Edward Lewis +1-703-227-9854 ARIN Research Engineer ...as graceful as a blindfolded bull in a china shop...
participants (3)
-
Daniel Karrenberg -
Edward Lewis -
Patrik Fältström