- db-wg - mailman.ripe.net

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

17 Oct '17

Hi Job I agree with your assessment and, although it should be in a separate thread later, I think it's worth having a discussion about notifying resource holders if a foreign ROUTE object is created in the RIPE Database. But lets focus on the main points for now. cheersdenis co-chair DB WG From: Job Snijders via db-wg <db-wg(a)ripe.net> To: den is <denis1(a)gmail.com> Cc: Database WG <db-wg(a)ripe.net> Sent: Tuesday, 17 October 2017, 10:06 Subject: Re: [db-wg] Foreign ROUTE objects in RIPE Database - final decision? Dear Denis, Just to make sure we are on the same page, the 2 items presented by Bill are not two mutually exclusive options, I think both need to be done. On Tue, Oct 17, 2017 at 01:04:08AM +0000, den is via db-wg wrote: > 3) Consider possible, simple options to allow non RIPE resource > holders to 'approve' (if not authorise) the creation of a foreign > ROUTE object. An even simpler option is to not even attempt to authorize the creation of non-RIPE route objects, but simply flag them as "RIPE-NONAUTH". If people want to create authoritive route objects for non-RIPE-managed space they can create those objects in the DB from the RIR where the space is managed (APNIC, AFRINIC, etc). While on the surface of things, it looks trivial to have the RIPE database software fire off an email to what most probably is the legitimate 'foreign' owner of the IP block, it introduces quite some complexity such as "what if the resource is transfer to a new owner" "what if the resource is transferred to a new owner and a new RIR". Even though at some point in the past a form of approval was given, a day later that approval may have been revoked. The RIPE DB would have to continuously track movement of IP space. The continuously tracking of ownership aspects in other databases certainly can be done, but now the project no longer is simple low hanging fruit. I see some value if the RIPE DB could notify owners of non-RIPE IP blocks when non-authoritative data is created in the RIPE DB, but I think that would be something to consider in the future. Kind regards, Job

1 1

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

17 Oct '17

Hi Randy I don't think anyone cares how much work a script has to do, whether it's the one that writes or reads the data. But keep in mind that the RIPE Database was designed to be, and still largely is, human readable. So for the poor humans who delve into the database manually it's probably better not to confuse them with extra, new tags on 95% of the data that has the same meaning as before, and just tag that little 5% (or so) that can now be interpreted differently. cheersdenisco-chair DB WG From: Randy Bush via db-wg <db-wg(a)ripe.net> To: job(a)instituut.net Cc: Nick Hilliard via db-wg <db-wg(a)ripe.net> Sent: Tuesday, 17 October 2017, 19:34 Subject: Re: [db-wg] Foreign ROUTE objects in RIPE Database - final decision? >>> 2) Flag "route:" objects for non-RIPE-managed space with "source: >>> RIPE-NONAUTH" to identify non-authoritative data. >> >> why not go the other, more positive, direction, and identify authorised >> data with RIPE-AUTH or whatever? no need to be pejorative. > > This is not meant to be pejorative. "RIPE-NONAUTH" was brought up > because that way we don't touch the majority of objects that actually > are valid, and belong with RIPE. ahh, the poor overworked script will have less to do so we will have to pay it less. good plan. > If you have a different suggestion than "RIPE-NONAUTH" I think the group > would be all ears. "RIPE-2", "RIPE-OTHER", all is fine with me. for those old enough to remember, RARE

1 0

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

17 Oct '17

3) Consider possible, simple options to allow non RIPE resource holders to 'approve' (if not authorise) the creation of a foreign ROUTE object. cheers denis co-chair DB WG On 17 October 2017 at 01:23, William Sylvester via db-wg <db-wg(a)ripe.net> wrote: > > > ---------- Forwarded message ---------- > From: William Sylvester <william.sylvester(a)addrex.net> > To: Database WG <db-wg(a)ripe.net> > Cc: > Bcc: > Date: Mon, 16 Oct 2017 23:22:55 +0000 > Subject: Re: [db-wg] Foreign ROUTE objects in RIPE Database - final decision? > Db-wg members, > > Now that there has been substantial discussion on this topic, I wanted to try to bring the conversation back to a more practical and actionable path forward. Clearly this is a topic where there are a lot of opinions. From seeing the discussion and receiving additional feedback, I wanted to point out some re-occurring themes and request a revised proposal for working group consideration. > > > 1) Remove the "origin:" authorization requirement. RIPE is currently the only RIR that requires this, the current implementation creates data concurrency issues across Internet databases by requiring the creation of duplicate autnums. > > 2) Flag "route:" objects for non-RIPE-managed space with "source: RIPE-NONAUTH" to identify non-authoritative data. > > > I’d like to seek input from the group; > > - Do you feel these options are straight forward and easy to understand? > - Does this provide a reasonable security posture? > - Will this make a positive impact for RIPE DB and the greater Internet? > - Do you support these concepts? > > > Thanks, > William > Db-wg co-chair > >

2 1

Re: [db-wg] violation of DB T&C by DB-WG co-chair, is it acceptable?
by Brian Nisbet 17 Oct '17

17 Oct '17

On 14/10/17 12:37, Carsten Schiefner wrote: >> William, >> >> On 13.10.2017 21:56, William Sylvester wrote: >>> [...] >>> >>> 2. Conduct of a co-chair for a working group. This is a much more >>> challenging issue, should a chair be held accountable for the >>> actions of an employer? >> >> are we really talking about an action by some random colleagues of David >> he'd barely know as they would work in another department across the >> street and therefore he'd almost never have anything to do with them? > > I can speak to this from personal experience. > > When I worked for BT, a misguided salesdroid mined the RIPE DB and > used it to send spam to possible customers. When I found out I > arranged for the salesdroid to be suitably formally warned, extracted > an apology from them and passed the apologies of the company to the > RIPE NCC and the community. At the time I was a co-chair of the DB WG. > > I would expect any member of the community (not just a WG chair) to do > likewise. > > Nigel I would very much like to see, at the very least, something similar to the above, very clearly and formally. There are, it should be pointed out, processes for removal of WG Chairs, however the DB WG has not decided to have one. There are examples out there that the WG could adopt if it wishes. The problems that occurred in Rome were one of the reasons the community felt that a process was a good idea, but remember, lack of documented process didn't stop things happening on that occasion. The WG has the power to effect change if needed. This is *still* not a great state of affairs and while I don't want to take away from the points that must be discussed in Dubai, I would, at this point, expect it to be on the agenda at the meeting. Thanks, Brian

1 0

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

17 Oct '17

Daniel Shaw via db-wg wrote: > On 17/10/2017, 03:24, William Sylvester via db-wg typed: >> > >> > >> > 1) Remove the "origin:" authorization requirement. RIPE is currently the only RIR that requires this, the current implementation creates data concurrency issues across Internet databases by requiring the creation of duplicate autnums. >> > >> > 2) Flag "route:" objects for non-RIPE-managed space with "source: RIPE-NONAUTH" to identify non-authoritative data. >> > >> > >> > I’d like to seek input from the group; >> > >> > - Do you feel these options are straight forward and easy to understand? >> > - Does this provide a reasonable security posture? >> > - Will this make a positive impact for RIPE DB and the greater Internet? >> > - Do you support these concepts? >> > > > Yes. There isn't really any downside to these. There are clear benefits to both. Agreed. This sounds like a sensible thing to do. Nick

1 0

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

17 Oct '17

On 17/10/2017, 03:24, William Sylvester via db-wg typed: > > > 1) Remove the "origin:" authorization requirement. RIPE is currently the only RIR that requires this, the current implementation creates data concurrency issues across Internet databases by requiring the creation of duplicate autnums. > > 2) Flag "route:" objects for non-RIPE-managed space with "source: RIPE-NONAUTH" to identify non-authoritative data. > > > I’d like to seek input from the group; > > - Do you feel these options are straight forward and easy to understand? > - Does this provide a reasonable security posture? > - Will this make a positive impact for RIPE DB and the greater Internet? > - Do you support these concepts? > Yes. There isn't really any downside to these. There are clear benefits to both. Regards, Daniel

1 0

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

17 Oct '17

Dear WG, chairs, On Mon, Oct 16, 2017 at 11:24:14PM +0000, William Sylvester via db-wg wrote: > Now that there has been substantial discussion on this topic, I wanted > to try to bring the conversation back to a more practical and > actionable path forward. Clearly this is a topic where there are a lot > of opinions. From seeing the discussion and receiving additional > feedback, I wanted to point out some re-occurring themes and request a > revised proposal for working group consideration. > > 1) Remove the "origin:" authorization requirement. RIPE is currently > the only RIR that requires this, the current implementation creates > data concurrency issues across Internet databases by requiring the > creation of duplicate autnums. > > 2) Flag "route:" objects for non-RIPE-managed space with "source: > RIPE-NONAUTH" to identify non-authoritative data. > > I’d like to seek input from the group; > > - Do you feel these options are straight forward and easy to understand? > - Does this provide a reasonable security posture? > - Will this make a positive impact for RIPE DB and the greater Internet? > - Do you support these concepts? I fully support executing both action items. Removal of the "origin:" authorisation step will bring the RIPE IRR in line with other databases such as APNIC, RADB and as such make use of the database easier for global organisations. It also aligns with the practises observed in RPKI. Flagging "route:/route6:" objects with "RIPE-NONAUTH" when they cover non-RIPE managed space has many benefits. Setting the "source:" to represent what the data actually means, we can programmatically analyse the contents of the RIPE DB and facilitate creation of higher quality route filters, but also for instance use it as a reputation factor in spam scoring. These two options represent what currently is the lowest hanging fruit to improve in this problem space. Funny enough, neither of these items makes things more difficult for anyone involved, on the contrary - by making route: authorisation easier, _and_ showing what is authoritative data and what is not, we improve the security posture of the RIPE community. Kind regards, Job

4 3

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

17 Oct '17

Dear Denis, Just to make sure we are on the same page, the 2 items presented by Bill are not two mutually exclusive options, I think both need to be done. On Tue, Oct 17, 2017 at 01:04:08AM +0000, den is via db-wg wrote: > 3) Consider possible, simple options to allow non RIPE resource > holders to 'approve' (if not authorise) the creation of a foreign > ROUTE object. An even simpler option is to not even attempt to authorize the creation of non-RIPE route objects, but simply flag them as "RIPE-NONAUTH". If people want to create authoritive route objects for non-RIPE-managed space they can create those objects in the DB from the RIR where the space is managed (APNIC, AFRINIC, etc). While on the surface of things, it looks trivial to have the RIPE database software fire off an email to what most probably is the legitimate 'foreign' owner of the IP block, it introduces quite some complexity such as "what if the resource is transfer to a new owner" "what if the resource is transferred to a new owner and a new RIR". Even though at some point in the past a form of approval was given, a day later that approval may have been revoked. The RIPE DB would have to continuously track movement of IP space. The continuously tracking of ownership aspects in other databases certainly can be done, but now the project no longer is simple low hanging fruit. I see some value if the RIPE DB could notify owners of non-RIPE IP blocks when non-authoritative data is created in the RIPE DB, but I think that would be something to consider in the future. Kind regards, Job

1 0

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

17 Oct '17

Hi, Op 17 okt. 2017, om 09:53 heeft Job Snijders via db-wg <db-wg(a)ripe.net> het volgende geschreven: > These two options represent what currently is the lowest hanging fruit > to improve in this problem space. Funny enough, neither of these items > makes things more difficult for anyone involved, on the contrary - by > making route: authorisation easier, _and_ showing what is authoritative > data and what is not, we improve the security posture of the RIPE > community. +1 This is low-hanging fruit indeed, and I see no problems with either of them. Cheers, Sander

1 0

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

17 Oct '17

On 11/10/2017, 18:22, Sascha typed: > > >> Question - Should the RIPE Database allow creation of ROUTE objects for non RIPE resources? This question has to be split into two to be able to be answered. Part 1: Should RIPE DB allow creation of route/route6 objects when both the inetnum(6) and aut-num resources are non-RIPE? Part 2: Should RIPE DB allow creation of route/route6 objects where one or the other of the prefix or origin asn *are* in the RIPE DB, and the other is not. That is so call cross-rir route objects? Part 2 is trickier, and largely what this thread is debating as I understand it. For part 1, I struggle to see any good reason to allow this to continue. It simple allows for anyone to create any route: objects, referring to any combination of resources that may or may not have any bearing on the reality of the internet. There is not even a good verification the resources are even allocated at all. In addition, once set up, there is zero incentive for anyone to ever delete these once obsolete. This was necessary once upon a time, as there were no useful alternatives then. That is now history. > > Is an option D: create a central IRRDB with authentication hooks > into all RIRs completely out of the question? I don't see that as a practical or realistic option. Regards, Daniel

1 0