- db-wg - mailman.ripe.net

Re: [db-wg] Foreign ROUTE objects in RIPE Database - final decision?
by Lu Heng 10 Oct '17

10 Oct '17

On Tue, Oct 10, 2017 at 18:33 Sander Steffann via db-wg <db-wg(a)ripe.net> wrote: > Hi, > > > In any case, given that we have no proper global registry, and we have > > lots of cross-region routes ("addresses from RIR A, AS number from RIR > B"), > > we need to find a way to document these properly. > > My preference would be to start such a global registry. I would love it if > at least RIPE, APNIC and AFRINIC (who all use very similar data models > IIRC) could combine databases. Having separate (but in reality partially > overlapping) databases for a global resource will always result in > discussions like we have had over the last two+ years IMHO. Let's make an > effort to form a single point of reference. > +1 > > Cheers, > Sander > > > -- -- Kind regards. Lu

1 0

Re: [db-wg] Foreign ROUTE objects in RIPE Database - final decision?
by Sander Steffann 10 Oct '17

10 Oct '17

Hi, > In any case, given that we have no proper global registry, and we have > lots of cross-region routes ("addresses from RIR A, AS number from RIR B"), > we need to find a way to document these properly. My preference would be to start such a global registry. I would love it if at least RIPE, APNIC and AFRINIC (who all use very similar data models IIRC) could combine databases. Having separate (but in reality partially overlapping) databases for a global resource will always result in discussions like we have had over the last two+ years IMHO. Let's make an effort to form a single point of reference. Cheers, Sander

1 0

Re: [db-wg] Foreign ROUTE objects in RIPE Database - final decision?
by Nick Hilliard 09 Oct '17

09 Oct '17

Clement Cavadore via db-wg wrote: > That one, but with a *strong* enhancement on a trust model for such > objects. We just cannot simply allow non-ripe route objects to be > inserted in the db without any legitimity checks. there is no way which non-ripe route objects can be subjected to legitimacy checks. It can't be done because several other RIRs don't provide the authentication hooks to be able to do this. On the question of this work item, what I would like is a mechanism to distinguish between route objects which have been verified using the trust grandfathering mechanism deployed by the RIPE NCC, and those which have not. There are many ways of doing this, e.g. 1. Option B: remove all non RIPE objects from the db. 2. use different source: tags to distinguish between ripe space and non ripe space. 3. leave whois.ripe.net as it is, and create a different database consisting of only authenticated route: objects. 4. leave whois.ripe.net as it is and create a different database view which, if queried, will return only objects which have been authenticated. Options for handling this might include: 4.1. same db on different hostname providing a different view 4.2. a new ripe-181++ tag identifying authenticated objects My preferences in order are: 2, then 4.2, then 4.1, then 3, then 1. No doubt there are plenty of other ways to crack this particular nut. Denis's description of Option C perpetuates what is a operational problem which has serious and ongoing consequences in the area of prefix hijacking. It would not be realistic to characterise the current situation as "if it ain't broke...", because it is identifiably broken, and it is this breakage which has led to this discussion. If we could solve this problem in a way which was operationally portable to other RIRs, that would be a bonus. Nick

1 0

Re: [db-wg] Foreign ROUTE objects in RIPE Database - final decision?
by Gert Doering 09 Oct '17

09 Oct '17

Hi, On Mon, Oct 09, 2017 at 03:48:43PM +0200, denis walker via db-wg wrote: > Question - Should the RIPE Database allow creation of ROUTE objects for non RIPE resources? I wonder if this is the right question to ask, or we should step back and ask the question what can we do to provide a secure routing registry where people can document "this prefix will be announced by that AS" in a way that is a) strongly authenticated (= only the current holder of the address space can create such a route: object), and b) easy to use for people building filters (= not having to walk 5 different IRRDBs to find the set of prefixes announced by a possible customer)? instead? A, B or C may be a consequence of this. In any case, given that we have no proper global registry, and we have lots of cross-region routes ("addresses from RIR A, AS number from RIR B"), we need to find a way to document these properly. From a user perspective, I like to have all route: objects for my customers in a single place (RIPE BD, that is), so for me, "A" would be most convenient - but we need to add some sort of cross-RIR authentication layer. Gert Doering -- NetMaster -- have you enabled IPv6 on something today...? SpaceNet AG Vorstand: Sebastian v. Bomhard Joseph-Dollinger-Bogen 14 Aufsichtsratsvors.: A. Grundner-Culemann D-80807 Muenchen HRB: 136055 (AG Muenchen) Tel: +49 (0)89/32356-444 USt-IdNr.: DE813185279

1 0

Re: [db-wg] Foreign ROUTE objects in RIPE Database - final decision?
by Hank Nussbacher 09 Oct '17

09 Oct '17

On 09/10/2017 18:22, denis walker via db-wg wrote: C -Hank

1 0

Re: Foreign ROUTE objects in RIPE Database - final decision?
by Alexander Stranzky 09 Oct '17

09 Oct '17

personal opinion: I don't care as a programmer. company's opinion: A - all our network resources are routed solely through our RIPE-AutNum. -- velia.net Internetdienste GmbH * Hessen-Homburg-Platz 1 * D-63452 Hanau Geschäftsführer: Franz G. Köhler, Arek Akilli * AG Hanau * HRB 7588 Tel: +49 6181 1898119 * http://www.velia.net Follow us on twitter: https://twitter.com/velia_net

1 0

Re: [db-wg] Foreign ROUTE objects in RIPE Database - final decision?
by Clement Cavadore 09 Oct '17

09 Oct '17

Hello Denis, WG On Mon, 2017-10-09 at 15:49 +0200, denis walker via db-wg wrote: > A: Yes the RIPE Database should allow creation of ROUTE objects for > non RIPE resources. If this is the answer then we can look at how to > make the data more trusting in the short term (and there may be ways > to do that). That one, but with a *strong* enhancement on a trust model for such objects. We just cannot simply allow non-ripe route objects to be inserted in the db without any legitimity checks. -- Clément Cavadore

1 0

Re: [db-wg] Foreign ROUTE objects in RIPE Database - final decision?
by Sascha Luck [ml] 09 Oct '17

09 Oct '17

Denis, On Mon, Oct 09, 2017 at 02:49:39PM +0100, denis walker via db-wg wrote: >Question - Should the RIPE Database allow creation of ROUTE >objects for non RIPE resources? > >A: Yes the RIPE Database should allow creation of ROUTE objects >for non RIPE resources. If this is the answer then we can look >at how to make the data more trusting in the short term (and >there may be ways to do that). > >B: No the RIPE Database should not allow creation of ROUTE >objects for any non RIPE resource. If this is the answer then >maybe we need to look at a project to remove all non RIPE >resource related ROUTE(6)/AUT-NUM objects from the RIPE Database >and disable the feature. But that needs to look at what to do >with ROUTE objects for prefixes from one region and ASNs from a >different region. > >C: Just leave things as they are. Don't promote the service but >no projects to remove any data from the RIPE Database either. >Just drift along on the principle "if it ain't broken, don't fix >it". This also means document this as the answer to this >question and no more discussions. Of course any problems >associated with this feature don't go away. Either A or C, insofar as they are different. Not allowing "foreign" resources in the ripedb denies the reality that there will always be RIPE-allocated prefixes originated from non-RIPE-assigned ASNs and vice versa. One could make an argument for not allowing route: objects where both the prefix and ASN are from outside RIPE. This would complicate (or make impossible) any attempt at automating acceptance filters, unless and until this information is available from other (trusted?) sources. As it currently stands, this would, on balance, be worse for the "good of the internet". cheers, Sascha Luck

1 0

Re: [db-wg] Foreign ROUTE objects in RIPE Database - final decision?
by Lu Heng 09 Oct '17

09 Oct '17

A On Mon, Oct 9, 2017 at 21:48 denis walker via db-wg <db-wg(a)ripe.net> wrote: > Colleagues > > This has been discussed many times with many views expressed. We are sure > lots of you are bored with commenting on this issue. Unfortunately no > consensus has yet been found. So we would like to take it back to the basic > question and see if we can get a conclusive answer. This is not a question > specifically about the Afrinic homing project. It's a more general question > about the RIPE Database providing this 'feature'. Even if you have > commented before, please do so once more so we can get a definitive answer > to this question. I know there is a tendency to say nothing if someone else > has made the point you support. But this time we do need numbers. So even > if what you are thinking has been said, please at least add a +1 so we know > you also agree with that point. > > Question - Should the RIPE Database allow creation of ROUTE objects for > non RIPE resources? > > We see three most likely answers to this question. Each possible answer > does have consequences. > > A: Yes the RIPE Database should allow creation of ROUTE objects for non > RIPE resources. If this is the answer then we can look at how to make the > data more trusting in the short term (and there may be ways to do that). > > B: No the RIPE Database should not allow creation of ROUTE objects for any > non RIPE resource. If this is the answer then maybe we need to look at a > project to remove all non RIPE resource related ROUTE(6)/AUT-NUM objects > from the RIPE Database and disable the feature. But that needs to look at > what to do with ROUTE objects for prefixes from one region and ASNs from a > different region. > > C: Just leave things as they are. Don't promote the service but no > projects to remove any data from the RIPE Database either. Just drift along > on the principle "if it ain't broken, don't fix it". This also means > document this as the answer to this question and no more discussions. Of > course any problems associated with this feature don't go away. > > > I will briefly review the discussions over the last 2 years: > > At the update given at RIPE 72 for the removal of the data starting with > the Afinic homing project, there was no consensus and a call for more > comments on the mailing list. A discussion continued on the mailing list > until October. There was both support and opposition to removing these > objects from the RIPE Database. One point raised during this discussion > regarding the Afrinic homing project was, "I believe that what is needed > here most of all is a clear message from both communities that it is > desirable to have the ROUTE(6) objects for AFRINIC space in the AFRINIC IRR > where the registered resource holders can properly authorise objects." No > one has yet referred to a declared consensus of opinion by the Afrinic > community to the removal of these objects from the RIPE Database. Also > during this discussion Afrinic said that they have not yet worked out all > the fine details of the proposed move of this data. Several issues were > raised on the technicalities of moving this data and it was said in the > discussion that no one has yet even done the analysis to see if any of > these conditions exist (like duplicate, but different data in both > databases). > > At the update given at RIPE 73, there was still no consensus. The RIPE NCC > and Afrinic were asked to do a check on what could go wrong and who members > could contact in the event of things going wrong. This doesn't seem to have > been done. The RIPE NCC suggested they may be able to produce a video or > webinar about the transfer. This doesn't seem to have been done either. The > call was again made that both the RIPE and Afrinic communities re-state > that they both want this project to go forward. That has not happened. It > was said that Afrinic is the first of many RIRs that may want data > transferred out of the RIPE Database. Is it the RIRs or the communities > that want this? The only question raised in the meeting was opposed to the > transfer. No discussion followed on the mailing list. > > At RIPE 74 there was no update on this issue. > > The next discussion wasn't until July 2017. In this discussion there was > mention of developing a global IRR database with a couple of supporters. We > all know such ideas often take years to be agreed on, designed and > implemented as a practical solution. We want to answer this question in > relation to the situation today. There were some pointers made back to > previous mailing list discussions, but nothing else new was said. > > cheers > denis > co-chair DB WG > -- -- Kind regards. Lu

1 0

Foreign ROUTE objects in RIPE Database - final decision?
by denis walker 09 Oct '17

09 Oct '17

Colleagues This has been discussed many times with many views expressed. We are sure lots of you are bored with commenting on this issue. Unfortunately no consensus has yet been found. So we would like to take it back to the basic question and see if we can get a conclusive answer. This is not a question specifically about the Afrinic homing project. It's a more general question about the RIPE Database providing this 'feature'. Even if you have commented before, please do so once more so we can get a definitive answer to this question. I know there is a tendency to say nothing if someone else has made the point you support. But this time we do need numbers. So even if what you are thinking has been said, please at least add a +1 so we know you also agree with that point. Question - Should the RIPE Database allow creation of ROUTE objects for non RIPE resources? We see three most likely answers to this question. Each possible answer does have consequences. A: Yes the RIPE Database should allow creation of ROUTE objects for non RIPE resources. If this is the answer then we can look at how to make the data more trusting in the short term (and there may be ways to do that). B: No the RIPE Database should not allow creation of ROUTE objects for any non RIPE resource. If this is the answer then maybe we need to look at a project to remove all non RIPE resource related ROUTE(6)/AUT-NUM objects from the RIPE Database and disable the feature. But that needs to look at what to do with ROUTE objects for prefixes from one region and ASNs from a different region. C: Just leave things as they are. Don't promote the service but no projects to remove any data from the RIPE Database either. Just drift along on the principle "if it ain't broken, don't fix it". This also means document this as the answer to this question and no more discussions. Of course any problems associated with this feature don't go away. I will briefly review the discussions over the last 2 years: At the update given at RIPE 72 for the removal of the data starting with the Afinic homing project, there was no consensus and a call for more comments on the mailing list. A discussion continued on the mailing list until October. There was both support and opposition to removing these objects from the RIPE Database. One point raised during this discussion regarding the Afrinic homing project was, "I believe that what is needed here most of all is a clear message from both communities that it is desirable to have the ROUTE(6) objects for AFRINIC space in the AFRINIC IRR where the registered resource holders can properly authorise objects." No one has yet referred to a declared consensus of opinion by the Afrinic community to the removal of these objects from the RIPE Database. Also during this discussion Afrinic said that they have not yet worked out all the fine details of the proposed move of this data. Several issues were raised on the technicalities of moving this data and it was said in the discussion that no one has yet even done the analysis to see if any of these conditions exist (like duplicate, but different data in both databases). At the update given at RIPE 73, there was still no consensus. The RIPE NCC and Afrinic were asked to do a check on what could go wrong and who members could contact in the event of things going wrong. This doesn't seem to have been done. The RIPE NCC suggested they may be able to produce a video or webinar about the transfer. This doesn't seem to have been done either. The call was again made that both the RIPE and Afrinic communities re-state that they both want this project to go forward. That has not happened. It was said that Afrinic is the first of many RIRs that may want data transferred out of the RIPE Database. Is it the RIRs or the communities that want this? The only question raised in the meeting was opposed to the transfer. No discussion followed on the mailing list. At RIPE 74 there was no update on this issue. The next discussion wasn't until July 2017. In this discussion there was mention of developing a global IRR database with a couple of supporters. We all know such ideas often take years to be agreed on, designed and implemented as a practical solution. We want to answer this question in relation to the situation today. There were some pointers made back to previous mailing list discussions, but nothing else new was said. cheers denis co-chair DB WG

1 0