RE: Deletion of German domains
According to the agreement between RIPE NCC and DENIC we plan
Has this agreement been published?
to delete German (.de) domain objects from RIPE whois database on June 28th. From then on, information about German domains will be available from DENIC's whois server, whois.nic.de.
While speed and efficiency are to be commended, this is rather short notice, isn't it? From the mail on 8th June, I understood that DENIC would need between one and two months after moving the domain objects before the associated person objects could be moved. Regards. Mike
mike.norris@heanet.ie wrote:
According to the agreement between RIPE NCC and DENIC we plan
Has this agreement been published?
No, this is a working agreement about technical details. It is based on the community consensus that domain related information (excluding reverse delegations) should be moved from the RIPE Database to the respective ccTLD databases. The RIPE Database will still be the source for domain related information using referral mechanism (as long as top level domain objects point to the actual domain databases). Though it may be more efficient to query these databases directly.
to delete German (.de) domain objects from RIPE whois database on June 28th. From then on, information about German domains will be available from DENIC's whois server, whois.nic.de.
While speed and efficiency are to be commended, this is rather short notice, isn't it? From the mail on 8th June, I understood that DENIC would need between one and two months after moving the domain objects before the associated person objects could be moved.
You are right, we are moving only domain objects. The person objects will remain in the RIPE Database for several months.
Regards.
Mike
Regards, Andrei Robachevsky DB Group Manager RIPE NCC
You are right, we are moving only domain objects. The person objects will remain in the RIPE Database for several months.
and how will person:s work? in general, i am confused about this. e.g. whois -h whois.ripe.net dom.ain causes a referral to a domain: object with a tech-c: of <handle>-RIPE, or is it <handle>-DENIC? if the former, <handle>-RIPE then it would seem that either o all the person:s stay at ripe database, or o the person:s are mirrored from denic database to ripe database if the latter, <handle>-DENIC, then it would seem that either o all the person:s are mirrored in the ripe database, or o the above whois query will not be able to show the requestor the person:s in general there is a problem of duplication (or lack thereof as ripe seems uninclined to mirror any data except radb and apnic) as oposed to referral for handles. i am trying to understand the best practice we can muster now and what we might do in the future. for the curious, i run the RGNET database at whois.rg.net. the purpose is twofold, referree for some cctlds and my routing data. radb mirrors these data but ripe does not. i am confused whether i am rb366 or rb366-arin, and in how many databases i need to maintain the person:s. randy
For the .is domains we did this all in one evening. We listed out all the object maintained by our MNT object and registered them all in our local database where the -RIPE part of the objects was changed to -IS and then added a refer record to the is domain in the ripe-db. After that we deleted all the subdomains in the ripe-db along with all the unreferenced objects. Since then everything has been working fine except for a few users that didn't get the nic-handle change the first time, but now it has been 2 months and it seems like it never was the other way around ;) Oli On Tue, 27 Jun 2000, Randy Bush wrote:
You are right, we are moving only domain objects. The person objects will remain in the RIPE Database for several months.
and how will person:s work? in general, i am confused about this. e.g.
whois -h whois.ripe.net dom.ain
causes a referral to a domain: object with a tech-c: of <handle>-RIPE, or is it <handle>-DENIC?
if the former, <handle>-RIPE then it would seem that either o all the person:s stay at ripe database, or o the person:s are mirrored from denic database to ripe database
if the latter, <handle>-DENIC, then it would seem that either o all the person:s are mirrored in the ripe database, or o the above whois query will not be able to show the requestor the person:s
in general there is a problem of duplication (or lack thereof as ripe seems uninclined to mirror any data except radb and apnic) as oposed to referral for handles. i am trying to understand the best practice we can muster now and what we might do in the future.
for the curious, i run the RGNET database at whois.rg.net. the purpose is twofold, referree for some cctlds and my routing data. radb mirrors these data but ripe does not. i am confused whether i am rb366 or rb366-arin, and in how many databases i need to maintain the person:s.
randy
-- Olafur Osvaldsson System Administrator Internet Iceland inc. Tel: +354 525-5291 Email: oli@isnet.is
roam.psg.com:/usr/home/randy> whois -h whois.ripe.net is domain: is descr: Top level domain for Iceland descr: Internet Iceland - INTIS descr: Taeknigardi descr: Dunhaga 5 descr: 107 Reykjavik descr: ICELAND admin-c: HJ18 tech-c: MO33 zone-c: MO33 nserver: isgate.is nserver: ns.eu.net nserver: sunic.sunet.se nserver: aos.brl.mil nserver: nisc.jvnc.net remarks: http://www.isnet.is/nic mnt-by: ISTLD-MNT mnt-lower: ISTLD-MNT changed: marius@isgate.is 19931123 changed: marius@isgate.is 19970728 source: IS person: Marius Olafsson address: IntIS/ISnet address: Taeknigardi address: Dunhaga 5 address: 107 Reykjavik address: ICELAND phone: +354 525 4747 phone: +354 551 9965 fax-no: +354 561 0999 e-mail: marius@isnet.is nic-hdl: MO33 mnt-by: ISTLD-MNT changed: marius@isnet.is 19950201 changed: marius@isnet.is 20000218 source: RIPE note the source:s. i.e. it looks to me as if you are maintining all person:s in two places. and it seems not to work as well as one might wish roam.psg.com:/usr/home/randy> whois -h whois.ripe.net isnet.is % Rights restricted by copyright. See http://www.ripe.net/ripencc/pub-services/db/copyright.html Referred query to host whois.isnet.is at port 43 timed out. Probable reason - far referral loop. To see the object stored in the RIPE database use the -R flag in your query. randy
Because the is tld has to exist in the ripe-db (for the "refer" option) the person objects need to exist in the ripe-db for that domain. So if you would lookup something like vortex.is in the ripe-db you can see that it has only person objects with the source IS and those objects are not in the ripe-db and can only be looked up in whois.isnet.is. So the objects within the is tld-record are allways going to be maintained in two places, but that are the only ones. About that error you got when looking up isnet.is I have never seen it before. I get about 1/3 of the queries on .is domains from joshua.ripe.net and I haven't got any complains about this before so it looks like it is working. And I have to say that since we set up this db we have run into quite a few delays in registering IP networks in the ripe-db that we could not have lived with when registering domains so now it is only ourselves to blaim if something is slow with the domains. Oli -- Olafur Osvaldsson System Administrator Internet Iceland inc. Tel: +354 525-5291 Email: oli@isnet.is
Because the is tld has to exist in the ripe-db (for the "refer" option) the person objects need to exist in the ripe-db for that domain. So if you would lookup something like vortex.is in the ripe-db you can see that it has only person objects with the source IS and those objects are not in the ripe-db and can only be looked up in whois.isnet.is.
without ripe mirroring IS, how can person:s be in in the ripe database with the source: IS? randy
Ok, maybe this didn't get out quite right, but they don't mirror the is-db and they don't need to, all the domains except the is tld are only in the is-db and thus the person objects only need to exist there, people should see the start of the whois query that is made to whois.ripe.net for .is domains and look the person objects up in the appropriate database, and if they do a query without -r they get the objects in the first query anyways. This is prepended to all .is queries in the ripe-db: The object shown below is NOT in the RIPE database. It has been obtained by querying a remote server: (whois.isnet.is) at port 43. To see the object stored in the RIPE database use the -R flag in your query. Oli On Tue, 27 Jun 2000, Randy Bush wrote:
without ripe mirroring IS, how can person:s be in in the ripe database with the source: IS?
randy
-- Olafur Osvaldsson System Administrator Internet Iceland inc. Tel: +354 525-5291 Email: oli@isnet.is
Randy Bush wrote:
You are right, we are moving only domain objects. The person objects will remain in the RIPE Database for several months.
and how will person:s work? in general, i am confused about this. e.g.
whois -h whois.ripe.net dom.ain
causes a referral to a domain: object with a tech-c: of <handle>-RIPE, or is it <handle>-DENIC?
<handle>-RIPE. This is a transition state which will last for several months.
if the former, <handle>-RIPE then it would seem that either o all the person:s stay at ripe database, or
All person objects will stay in the RIPE Database until moved to DENIC one (with different suffix).
o the person:s are mirrored from denic database to ripe database
if the latter, <handle>-DENIC, then it would seem that either o all the person:s are mirrored in the ripe database, or o the above whois query will not be able to show the requestor the person:s
in general there is a problem of duplication (or lack thereof as ripe seems uninclined to mirror any data except radb and apnic) as oposed to referral for handles. i am trying to understand the best practice we can muster now and what we might do in the future.
Current setup of the RIPE database requires that the database (source) should be self consistent - that is no external references are allowed. Referral mechanism for domains is different. I agree, duplication of handles is a problem, but so far it cannot be solved without violating referential integrity.
for the curious, i run the RGNET database at whois.rg.net. the purpose is twofold, referree for some cctlds and my routing data. radb mirrors these data but ripe does not. i am confused whether i am rb366 or rb366-arin, and in how many databases i need to maintain the person:s.
randy
Reagrds, Andrei Robachevsky DB Group Manager RIPE NCC
believe me, i do understand and have sympathy with the view consistancy issue. but ... please explain how i can reconcile
All person objects will stay in the RIPE Database until moved to DENIC one (with different suffix).
with
Current setup of the RIPE database requires that the database (source) should be self consistent - that is no external references are allowed. Referral mechanism for domains is different.
given whois -h whois.ripe.net dom.ain.de returning <handle>-DENIC, so i then follow with whois -h whois.ripe.net <handle>-DENIC i can only conclude, givem the first quoted paragraph above, that the data must be manually entered in both databases or that ripe mirrors denic's database. randy, clueless in seattle
Randy Bush wrote:
believe me, i do understand and have sympathy with the view consistancy issue. but ...
please explain how i can reconcile
All person objects will stay in the RIPE Database until moved to DENIC one (with different suffix).
with
Current setup of the RIPE database requires that the database (source) should be self consistent - that is no external references are allowed. Referral mechanism for domains is different.
given
whois -h whois.ripe.net dom.ain.de
returning <handle>-DENIC, so i then follow with
whois -h whois.ripe.net <handle>-DENIC
i can only conclude, givem the first quoted paragraph above, that the data must be manually entered in both databases or that ripe mirrors denic's database.
randy, clueless in seattle
During the transition phase scenario is slightly different: whois -h whois.ripe.net dom.ain.de will return <handle>-RIPE, though with source DENIC, which does not look very nice, I agree. Anyway there is an indication that "The object shown below is NOT in the RIPE database. It has been obtained by querying a remote server". This also means that during this phase DENIC mirrors the RIPE Database. When finally DENIC has its own person database one will get <handle>-DENIC indeed. But then there is no need to query whois.ripe.net for this handle. Sorry if I missed something. Regards, Andrei, waiting for comments in Amsterdam.
When finally DENIC has its own person database one will get <handle>-DENIC indeed. But then there is no need to query whois.ripe.net for this handle.
i.e. in the long run we have to teach people to query <number of cctld> separate databases? this is not reasonable. randy
On Tue, Jun 27, 2000 at 04:07:00PM -0700, Randy Bush wrote:
When finally DENIC has its own person database one will get <handle>-DENIC indeed. But then there is no need to query whois.ripe.net for this handle. i.e. in the long run we have to teach people to query <number of cctld> separate databases? this is not reasonable.
I second that. That'll bring up lots of confusion and inconvenience at large. Udo
On Tue, Jun 27, 2000 at 04:07:00PM -0700, Randy Bush wrote:
When finally DENIC has its own person database one will get <handle>-DENIC indeed. But then there is no need to query whois.ripe.net for this handle.
i.e. in the long run we have to teach people to query <number of cctld> separate databases? this is not reasonable.
Of course it is reasonable, but not very convenient. In any case, we already have to query lots of seperate databases with lots of incompatible data formats. Just look at the gTLD registries, CH-NIC, UK Nominet or the asian registries. This is not a real problem. The problems arise whenever this non-structure changes and you have to adapt to yet another method of retrieving data. Unfortunately, DENIC is no exception to that rule. -- i.A. Michael van Elst / phone: +49 721 9652 330 Xlink - Network Information Centre \/ fax: +49 721 9652 349 Emmy-Noether-Strasse 9 /\ link http://nic.xlink.net/ D-76131 Karlsruhe, Germany /_______ email: hostmaster@xlink.net [ KPNQwest Germany GmbH, Sitz Karlsruhe ] [ Amtsgericht Karlsruhe HRB 8161, Geschaeftsfuehrer: Koen Bertoen ]
Quoting Michael van Elst <mlelstv@xlink.net>:
On Tue, Jun 27, 2000 at 04:07:00PM -0700, Randy Bush wrote:
When finally DENIC has its own person database one will get <handle>-DENI C indeed. But then there is no need to query whois.ripe.net for this handle .
i.e. in the long run we have to teach people to query <number of cctld> separate databases? this is not reasonable.
Of course it is reasonable, but not very convenient.
Several times different proposals have been made to alleviate this problem (without necessarily solving that of incomaptible data formats). What comes to mind is the use of SRV RRs in the TLDs (with appropriate heuristics to catch cases where actual registration takes place at a lower level, e.g. "co.uk"). Time to write up an I-D? -Peter
given
whois -h whois.ripe.net dom.ain.de
returning <handle>-DENIC, so i then follow with
whois -h whois.ripe.net <handle>-DENIC
i can only conclude, givem the first quoted paragraph above, that the data must be manually entered in both databases or that ripe mirrors denic's database.
Seems like we need a referal meganisme for nic-handles? Regards, Andre
On Tue, Jun 27, 2000 at 10:20:36PM +0200, Andre Koopal wrote:
given
whois -h whois.ripe.net dom.ain.de
returning <handle>-DENIC, so i then follow with
whois -h whois.ripe.net <handle>-DENIC
i can only conclude, givem the first quoted paragraph above, that the data must be manually entered in both databases or that ripe mirrors denic's database.
Seems like we need a referal meganisme for nic-handles?
You would need global nic-handles that somehow identify their database. Unfortunately this does not work if there is no single handle namespace. Fortunately this isn't really a problem as I (we?) don't want handle cross references. Such cross references only make sense if you want to achieve that a single person has a unique handle in all databases. While a unique and single handle for a person is a nice looking idea it simply does not work as maintainership of handles becomes more complex with the increasing number of database maintainers involved. When the person data is stored together with the domain data the handles are again only locally valid and a referral of the domain also resolves the person data nicely. There is no need for more complex mechanisms. BTW, those that really want cross references can translate whois data into hypertext and use URLs to locate data on arbitrary servers. -- i.A. Michael van Elst / phone: +49 721 9652 330 Xlink - Network Information Centre \/ fax: +49 721 9652 349 Emmy-Noether-Strasse 9 /\ link http://nic.xlink.net/ D-76131 Karlsruhe, Germany /_______ email: hostmaster@xlink.net [ KPNQwest Germany GmbH, Sitz Karlsruhe ] [ Amtsgericht Karlsruhe HRB 8161, Geschaeftsfuehrer: Koen Bertoen ]
On Wed, Jun 28, 2000 at 11:04:27AM +0200, Michael van Elst wrote:
On Tue, Jun 27, 2000 at 10:20:36PM +0200, Andre Koopal wrote:
given
whois -h whois.ripe.net dom.ain.de
returning <handle>-DENIC, so i then follow with
whois -h whois.ripe.net <handle>-DENIC
i can only conclude, givem the first quoted paragraph above, that the data must be manually entered in both databases or that ripe mirrors denic's database.
Seems like we need a referal meganisme for nic-handles?
You would need global nic-handles that somehow identify their database. Unfortunately this does not work if there is no single handle namespace.
All nic-hdls we are talking about seems to end in the name of their source: -RIPE, -IS, -DENIC (or -DE?) so in my opinion that makes them both unique and gives an easy way to build a referal meganism. The only handles that fails this are the internic hdls, but they have always been separate ... And we don't really want cross references for nic-hdls, although it might have the advantage of a single place to maintain your handle, but we want to extend the referal meganism in place for domains to work with handles as well. All IMHO. Regards, Andre
Fortunately this isn't really a problem as I (we?) don't want handle cross references. Such cross references only make sense if you want to achieve that a single person has a unique handle in all databases.
While a unique and single handle for a person is a nice looking idea it simply does not work as maintainership of handles becomes more complex with the increasing number of database maintainers involved.
When the person data is stored together with the domain data the handles are again only locally valid and a referral of the domain also resolves the person data nicely. There is no need for more complex mechanisms.
BTW, those that really want cross references can translate whois data into hypertext and use URLs to locate data on arbitrary servers.
-- i.A. Michael van Elst / phone: +49 721 9652 330 Xlink - Network Information Centre \/ fax: +49 721 9652 349 Emmy-Noether-Strasse 9 /\ link http://nic.xlink.net/ D-76131 Karlsruhe, Germany /_______ email: hostmaster@xlink.net [ KPNQwest Germany GmbH, Sitz Karlsruhe ] [ Amtsgericht Karlsruhe HRB 8161, Geschaeftsfuehrer: Koen Bertoen ]
[ note cc:s chopped to just wg ]
According to the agreement between RIPE NCC and DENIC we plan to delete German (.de) domain objects from RIPE whois database on June 28th. From then on, information about German domains will be available from DENIC's whois server, whois.nic.de. While speed and efficiency are to be commended, this is rather short notice, isn't it? From the mail on 8th June, I understood that DENIC would need between one and two months after moving the domain objects before the associated person objects could be moved.
given the referral system, why would the user know/care? i am using that referral system and hope i do not misunderstand it. randy
participants (8)
-
Andre Koopal -
Andrei Robachevsky -
Michael van Elst -
mike.norris＠heanet.ie -
Olafur Osvaldsson -
Peter Koch -
Randy Bush -
Udo Steinegger