[06:50:00] Geoff Huston joins the room [06:53:33] ggm joins the room [06:57:38] kurtis joins the room [06:58:43] dhask joins the room [06:59:02] Geoff Huston leaves the room [06:59:14] gih joins the room [06:59:21] roque joins the room [07:00:07] Doug_Otis joins the room [07:01:42] bje joins the room [07:01:42] dhask leaves the room [07:02:44] tachibana@jabber.org joins the room [07:03:02] rhe joins the room [07:03:09] haa joins the room [07:03:09] Brenden Kuerbis joins the room [07:05:05] weiler joins the room [07:05:26] about 60 people in the room. [07:05:45] Robert Kisteleki joins the room [07:05:49] David Cooper joins the room [07:05:55] randy joins the room [07:05:56] Hari joins the room [07:06:00] mlepinski joins the room [07:06:07] vijay.gill joins the room [07:06:17] sidrerrific [07:06:36] ChrisB joins the room [07:06:39] weiler leaves the room [07:06:53] shinmiyakawa joins the room [07:07:05] insider inside her insides [07:07:08] mlepinski leaves the room: Computer went to sleep [07:07:13] (the spider, inside her) [07:07:26] weiler joins the room [07:07:38] Matt Lepinski [07:07:47] sidr-arch, then roa-format [07:08:18] previously: digest and sig alg specified separately in each of 4+ drafts. moving that all to the certificate policy draft [07:08:28] psavola joins the room [07:08:40] dougm.home joins the room [07:09:55] Merridew joins the room [07:10:02] alg descrip in roa format [07:10:16] ggm leaves the room [07:10:28] Russ Housley [07:10:37] ggm joins the room [07:10:51] ggm leaves the room [07:11:22] many WGs have made this mistake: in 1999 smime had alg stuff scattered, they had to change whole suite of doc. don't want it in cert policy doc. want to change alg w/o blocking progression of thing sup standards track [07:11:47] in ipsec, we've said "here's the next-likely-to-be-MUST" [07:11:51] a SHOULD+ alg. [07:12:47] matt: you're recommending that the WG adopt something.... referenced by all 5? that would say the must algs are... the should+'s are.... ? Russ: right [07:13:01] sandy: clarifying above. [07:14:15] Michael Behringer joins the room [07:14:20] rob A: problem putting alg in CP: CP makes eyes glaze over, and i'm not sure implementers will read it. No problem w/ a signle doc, but it should not be the CP [07:14:57] Roque (our minutes scribe): [I didn't follow] [07:15:11] Matt: roa-format changes. [07:15:34] make xplicit that it can carry an arbitrary set of prefixes, including overlapping. [07:15:35] Jan Boogman joins the room [07:15:39] ggm joins the room [07:15:52] wants WG conset to alg removal. [07:15:56] consent, rather [07:16:03] is audio good? [07:16:12] yes, it's good here in australia [07:16:18] aalain joins the room [07:16:22] ROgue has said that separate document for algos will be better concerning diferent national situations with different algorithms [07:16:33] ASHIDA joins the room [07:16:52] Linus Nordberg joins the room [07:16:58] sidr arch: subject names: RIR/NIR not special cased. [07:17:11] ggm leaves the room [07:17:13] removed an example on resource xfer [07:17:17] pssara joins the room [07:18:07] sean.s.shen joins the room [07:18:45] subject name clarification: not meant to convey identity, uniqueness concerns [07:18:59] no known open issues; needs more review [07:19:23] yoshihiko.saeki@jabber.org joins the room [07:19:44] me: is IANA special-cased? No. [07:20:17] steve kent: CP revisions. [07:20:39] ggm joins the room [07:20:53] andrei Robachevsky is coordinating changes from RIRs. [07:21:06] most changes incorporated. [07:21:24] terminology changed globally. removed refs to rtg security. [07:22:31] going through terminology changes. [07:22:51] details on the slides. [07:24:17] [not repeating material from slides] [07:24:53] may want to come back to the CP approval procedures in the future. [07:25:37] Randy: what is called out? [07:25:58] kent: orgs administering CP will manage the change procedure [07:26:02] he said RIRs [07:26:08] randy laughed [07:26:15] laughed? [07:26:21] lellel joins the room [07:26:21] slide 6. [07:27:25] sandy (not as chair): is cust's responsiblity to renew entirely under their control? [07:27:34] bje leaves the room [07:27:38] kent: [legalese] [07:27:47] slide 7 [07:28:10] details re: contracts removed. [07:30:02] Aleksi Suhonen joins the room [07:30:02] open issues [07:30:20] where will you find CP? is it enough to say "RFC"? [07:30:23] kurtis leaves the room [07:30:36] url in topic gives me 404 [07:30:47] no extra charge [07:31:01] i guess it's from san francisco? [07:31:08] we said there' sno verified subscriber data. But what about SIA? we could modify the text to say "except SIA" [07:31:21] SIA is a heuristic. [07:31:52] ChrisB leaves the room [07:32:03] Aleksi Suhonen has set the subject to: SIDR WG http://www.ietf.org/proceedings/75/agenda/sidr.html [07:32:06] who can request revocation? should we reference CPS? [07:32:26] certainly the subject. what about others? the issuer? [07:32:43] would issuer entain 3rd party requests? defer to CPS. [07:33:38] where should alg specs reside? refer to previous discussion [07:33:38] mlepinski joins the room [07:33:55] housley. [07:34:14] re: who approves changes. slide 6. [07:34:51] russ got impression where local RIR/ISP policy have impact belong is CPS. so what's left in the CP MUST be implemented across the whole net. [07:35:07] that characterization defines the split. (one unknown clap) [07:35:21] ggm leaves the room [07:35:29] does this belong as BCP rather than STD? not sure what it means to advance up standards track. [07:35:38] randy: thanks to Russ. [07:35:41] ggm joins the room [07:35:46] randy; [HUH] [07:35:59] juampe.cerezo@gmail.com joins the room [07:36:00] ggm leaves the room [07:36:38] .... kent: could say RPKI signed objects are RFC-defined objects. [07:36:45] randy: likes that. doesn't want it open-ended. [07:37:11] juampe.cerezo@gmail.com leaves the room [07:37:12] randy: slide 5. kent: does not refer to NRO at all. [07:37:24] jpc joins the room [07:37:40] change control: Russ raised it. [07:37:49] Randy feels like a whacker. [07:38:00] Andrei R: [07:38:21] though these changes were reviewed by RIRs, not necessarily a common position. [07:38:29] randy: thanks to RIRs and Andrei [07:38:48] moving on..... [07:38:54] aalain leaves the room [07:39:19] Robert Kisteleki [07:39:41] rpsl sigs [07:40:10] changes in last ver: requires URL safenness [07:40:18] dropped case folding. [07:40:26] added tab-to-space conversion [07:40:41] ggm joins the room [07:41:05] [not as scribe] does anyone intend to really use this? [07:41:34] lipstick on a pig [07:41:38] more changes: clarified that EE cert must contain some resource relevant to this object [07:42:11] validity time clarification. see slide 4 of 6 [07:42:37] ggm is now known as not-ggm [07:42:38] not-ggm is now known as ggm [07:42:49] rainbow flags in the city has some influence on the minds ;) [07:43:11] merridew: eh? [07:43:25] larryblunk joins the room [07:43:30] "lipstik on the pig" issue ;) [07:43:51] kent: concerned .... anything being signed by this need to be related... need a strong alg specification. [07:44:03] robert: I tried. [07:44:20] kent: I'm uncomfortable w/ ahving anything signed which is orthogonal to cert. [07:44:47] there's nothing in the context of the issuance and validation of cert that gives you any reason to believe these assertions. [07:45:17] .... potential to undermine confidence in RPKI. might nto be consistent w/ CP. since it goes beyond attestation of resource holdings. [07:45:20] robert: disagree. [07:45:42] ggm is now known as not-ggm [07:45:42] not-ggm is now known as ggm [07:45:45] example re: contact info [07:46:00] It may be useful in a later rev of this draft to describe how RPSS-concepts would apply in the context of signed IRR objects [07:46:03] kent: you're talking about an application that will violate the CP. [07:46:31] kent: you're giving a licence to subject... to make arbitrary assertions... that's why we have CPs. [07:46:43] randy: agrees with kent [07:46:54] Robert Kisteleki leaves the room [07:47:34] Robert Kisteleki joins the room [07:47:38] [not as scribe] Rob: good job resisting the CP police. [07:48:39] Randy presenting. Dave will be after. [07:48:55] lebobits joins the room [07:49:08] Is there a presentation pack [07:49:15] preso available ? [07:49:33] COuld someone please ask the presenter wher the pack is [07:49:50] Sandy's working on it. [07:50:02] This is very difficult to follow without the preso [07:50:10] we know. [07:50:18] archive.psg.com/090730.sidr-rpki.pdf [07:50:27] ...was on the first slide, try that... [07:50:28] thx [07:50:42] You step in the stream, but the water has moved on. This page is not here. [07:50:42] ggm is now known as not-ggm [07:50:42] not-ggm is now known as ggm [07:51:05] I [07:51:40] Sorry, I must have noted it wrongly. :( [07:52:22] I'm sorry - but this is really quite difficult to follow without some kind of material to back up the descriptive words of "here" and "there", etc [07:53:06] sorry [07:53:23] not your fault sam [07:53:24] allocation in reaility slide: I break up my addr space in a variety of ways. [07:53:42] picture that looks like a radio spectrum chart [07:53:54] slide: fourth protocol. [07:55:10] showing objects followinf from net through rcynic gatherer to cache/server via rpi/rtr protocol to rtr [07:55:21] not a picture of message flows. [07:55:25] now, rather [07:55:34] http://www.ietf.org/proceedings/75/slides/sidr-7.pdf [07:55:42] ggm is now known as not-ggm [07:55:42] not-ggm is now known as ggm [07:55:49] lebobits leaves the room [07:55:49] lebobits joins the room [07:55:54] slide 12 [07:56:04] rababy joins the room [07:56:09] thanks for putting slides up [07:56:09] packet format slide [07:56:18] v4 prefix slide [07:56:29] for cache responses. [07:57:11] includes a random field 'cause someone wanted it. [07:57:21] questions? [07:57:21] lebo joins the room [07:57:50] dave ward. [07:59:04] wants to point out ho easy validation is [07:59:06] how [07:59:26] prefix validation logic slide [07:59:43] there is an error in the pseudocode. initilize prefix_exists to FALSE [08:00:43] ggm is now known as not-ggm [08:00:43] not-ggm is now known as ggm [08:00:58] debate about how to handle prefix outside the maxlen [08:01:13] bestpath selection [08:03:47] kurtis joins the room [08:03:54] [sorry for dropping so much] [08:04:29] requesting adoption [08:05:03] how do to selection alg will be in this doc. draft-huston might or mihgt not remain: ask Geoff. [08:05:29] Roque: asking for adoption of only this draft, or also protocol draft? [08:05:36] Ward: ask randy [08:05:43] ggm is now known as not-ggm [08:05:49] ward: we can adopt this in isolation, if we want. [08:06:05] lepinki: this is helpful. it should be adopted. [08:06:14] sandy: this to list, along w/ alg docs. [08:06:19] terry manderson. [08:06:53] on the usecases doc [08:07:03] projection problems [08:07:31] blames dave ward for inspiring this. [08:08:05] Michael Behringer leaves the room: Replaced by new connection [08:08:06] Michael Behringer joins the room [08:08:34] purposes: to see if we have a complete solution, focus on issues, develop test cases, and allow newcomers to comprehend the problem space. [08:09:50] goal: list all cases currently observed, previously see, possible, without bias [08:10:04] trying not to censor selves prematurely [08:10:25] four sections: origination, adjacency (not in charter so far), partial deployment, rep,ying party [08:10:31] relying party [08:10:44] not-ggm is now known as ggm [08:11:29] half a dozen or so read the draft. [08:11:40] which means most of the 60+ people in the room haven't. [08:12:08] Michael Behringer leaves the room: Replaced by new connection [08:12:09] Michael Behringer joins the room [08:12:22] 100+ I think [08:12:35] it was 60ish when we started. haven't counted since. [08:13:08] reminder about adjacency not being in charter now. [08:13:37] Q: have we missed anything?> does something not make sense? adopt now? or wait for a rev or two? [08:13:52] Danny Mcpherson: [08:14:18] how far do we have to go down design/deployment/etc before we start worrying about path validation. [08:14:31] sandy: I'd like to see some of the docs reach publication before we tackle such a big topic [08:14:37] danny; but it's related. [08:14:55] afriad we'll do thing s that constrain paqth validation. if we wait too long, some thing osified... [08:15:20] rob austein: this draft is not an ocean-boiling exercise, but t is a temperature rasing device. [08:15:29] larryblunk leaves the room [08:15:37] seems odd that path vlaidation is excluded. if we're going to raise the ave temp of the ocean, perhaps we can include that. [08:15:44] ggm is now known as not-ggm [08:15:44] not-ggm is now known as ggm [08:16:33] ggm is now known as ggm [08:16:34] ggm is now known as not-ggm [08:16:34] not-ggm is now known as ggm [08:17:01] ~80+ [08:17:11] Ross Callon: when this was set up, we didn't want to repeat the fun from RPSEC. if there's broad consensus that we want to work on path validation, .... we could put it into chater. we didn't put it in because we thought we could make progress on what was in the charter.... [08:17:42] ross: if chairs are going to discuss w/ ADs, nice to have sense of room. [08:17:52] sandy; finish queues before goig to that [08:18:02] The issue at this point is that RPSEC has not been able to generate a clear set of requirements in the area of path validation and the AD's concern about re-runing the RPSEC debate is a valid concern [08:18:25] randy: in pittsburgh in 2001?, bellovin/randy/kent went to dinner... "would you get this ath validation thing going". underlying that was the problem of attesting to resource ownership. [08:18:54] randy'; the fact that we're making progress shoudl not be a reason for confounding the situation now. the fact that we have it going to rtrs is a major win. let's not open the swamp. [08:19:13] 1st: ownership 2: origin validation 3: path validation 4: seeing that packets follow control plane. [08:19:34] if you missed iesg: by measurement, DFZ is 70% default. we ain't got the control plane. [08:19:38] let us finish THIS. [08:19:42] housley, as ietf chair. [08:20:31] charter says "when reqs come from rpsec, you can go on". ask the ADs where that's coming from. that problem is in the hands of the rtg ads. [08:20:43] ross: maybe I'm feeling better about not re-upping as rtg AD. [08:20:45] ggm is now known as ggm [08:20:45] ggm is now known as not-ggm [08:20:45] not-ggm is now known as ggm [08:20:57] ross: dont' get bogged down at expense of finishing what you're doing. [08:21:32] aalain joins the room [08:22:05] danny: [08:22:35] Hari leaves the room: Replaced by new connection [08:22:36] Hari joins the room [08:22:44] mlepinski leaves the room [08:23:11] path validation has to be considered at some point. would hate to see this constrain what we do w/ path validation [08:23:22] mlepinski joins the room [08:24:08] sandy: you can't do path validation w/o doing origin validation [08:25:03] paul hoffman: you said you didn't deal w/ adjacney because of charter, but you have an info doc... i think it's ok if there's a giant elephant in the room, that it be discussed in an info doc. [08:25:22] that might help, by describing/listing issues [08:25:45] ggm is now known as not-ggm [08:25:45] not-ggm is now known as ggm [08:25:45] ggm is now known as ggm [08:26:24] larry blunk: we keep talking about origin, but it's also prefix length. [08:26:49] that can be powerful. if attacker can only announce same length as your prefix, that could limit the scope of their attack. [08:27:00] some take over for a sec? [08:27:23] < ggm> thanks for scribing thus far sam [08:27:27] < ggm> most helpful [08:28:29] i'll be back [08:29:46] Hari leaves the room: Replaced by new connection [08:29:46] Hari joins the room [08:29:55] Ruediger: Whether it is routing or policy depends on what we're going to check. We don't know yet, which is why doing path validation isa very big topic. [08:30:06] thomas beckhaus joins the room [08:30:37] Sam Weiler: The IETf isn't a top-down organisation, if there are people that want to do the work, they should be able to, but it shouldn't dilute the work here, spin up another group to do it. [08:30:45] ggm is now known as not-ggm [08:30:45] not-ggm is now known as ggm [08:30:45] ggm is now known as ggm [08:30:48] i'm back [08:30:51] ASHIDA leaves the room [08:30:52] greg lebowitz: [08:30:55] Russ: I agree, it is in the lap of the routing ADs. [08:31:21] the thought I have when russ made his statement: you can do a lot of work on i-d's before they come into the wg. [08:32:03] greg: work on these in i-d's for now.. Russ's statement doesn't have to stop people from working on stuff. [08:32:27] sandy: plea: finish topics on the plate first. [08:32:35] greg: don't let it DOS the current charter [08:32:36] thomas beckhaus leaves the room [08:32:50] sandy: it's rife this opportunities to DOS anything it runs into [08:33:21] randy: we're in the middle of the DOS attack right now. 2) rpki work was done outside the ietf then it was brought into the WG when it was understood what was happening and we had running code [08:33:39] randy's shirt said "running code, and run away before someone comes to help" [08:34:11] sandy; incremental deployment is hard. please pay attention to those use cases. [08:34:27] think about signers and relying parties... [08:34:40] [me to ggm: most welcome.] [08:34:52] sriram. [08:35:18] from NIST. worked w terry et al on use case doc. [08:35:37] Linus Nordberg leaves the room [08:35:38] came up w/ problem interpreting ROAs. [08:35:48] ggm is now known as ggm [08:35:48] ggm is now known as not-ggm [08:36:06] AS A has a roa: prefix/19, max len 20 [08:36:53] AS B has EE cert for /24, but has created no ROA. B wants to multihome, announcing the /24. [08:37:13] randy; invalid. There is a covering roa. [08:37:23] you're announcing from a different AS. that's invalid. [08:37:34] terry: not by definition. draft don't say that. [08:37:37] randy: fix the drafts [08:37:49] terry: is the draft a WG doc? [08:37:54] yao joins the room [08:38:12] sriram: Q is: will that announcement be invalidation based on the AS A roa? [08:38:16] invalidated, rather [08:38:22] I should note I asked the WG to resolve this some months ago [08:38:45] andy: you're confused about maxlen. it says nothing about ownership. [08:38:45] I received 1 comment pro and 1 against a change in the ROA validation document [08:39:13] randy: what counts is "i own that 19" "i've not authed anyone to announce it, unless there's another ROA" [08:39:30] randy: isn't that what this is meant to defend against? the fact that there's an EE is irrelevant. [08:39:31] It would be good to understand the WG consensus to a slightly greater extent than just 2 comments. [08:39:32] rob a: [08:39:44] not covered in clear text in drafts. [08:39:50] but it is covered if you dig. [08:40:30] rob: you have the manifest. you know EVERY ROA that these prefixes have authorized. this is an attack. [08:40:34] not clearly stated, but it's there. [08:40:47] not-ggm is now known as ggm [08:40:47] ggm is now known as ggm [08:40:47] ggm is now known as not-ggm [08:40:49] Ruediger: if you want it accepted, you need another ROA. [08:41:05] Sandy: w/ data structures here, .... [08:41:11] is this accepted or not. [08:42:34] scudder: i think the answer is invalid. what larry said is illuminating.... [08:42:45] larry's example is helpful. [08:43:19] someone take over again?> [08:44:03] haa leaves the room [08:44:44] Matt: As editor of ROA format, I will try to make it clearer that max length is just a macro. [08:45:04] This discussion is evidence that we need a doc like dward described earlier. [08:45:29] Sandy: Agreed, what Terry was talking about earlier, what we're talking about now is validation. [08:45:32] draft-ietf-sidr-roa-vbalidation was PRECISELY the document Matt just asked for [08:45:47] i.e. the WG alkready adopted such a document [08:45:48] not-ggm is now known as ggm [08:45:48] ggm is now known as ggm [08:45:48] ggm is now known as not-ggm [08:46:28] and the discussion is actually about the contents of the existing WG document, not whether such a working group document should exist or not. [08:47:55] mlepinski leaves the room: Replaced by new connection [08:47:55] mlepinski joins the room [08:49:23] rhe leaves the room [08:49:41] rhe joins the room [08:50:05] steve kent: [08:50:12] local mgmt of trust anchors [08:50:31] it MAY help us w/ the contention over trust anchors. [08:50:48] not-ggm is now known as ggm [08:50:48] ggm is now known as ggm [08:50:48] ggm is now known as not-ggm [08:51:12] almost no good examples ofr tool for relying parties to enable local choice. [08:51:26] not-ggm leaves the room [08:51:33] yao leaves the room [08:52:03] we're makng life easy for rp: there's one TA, and he's it. [08:52:38] this motivates creating a tool to help RPs manage TA [08:52:49] the RP as the TA..... [08:53:01] in the GENERAL PKI case, not rpki..... [08:53:13] RP is on top [08:53:16] this is easy to do. [08:53:45] process: issue self-signed cert. acquire certs from A&B ad verify them [08:54:16] extract data from each cert, can modify extensions to match your policy. issue new A & B certs to match your policy [08:54:47] the RPKI case..... [08:55:15] self-signed RP cert must cover ALL addresses and ASNs [08:55:23] in the 3779 extensions [08:55:49] ggm joins the room [08:55:49] ggm is now known as not-ggm [08:55:49] not-ggm is now known as ggm [08:55:57] can subtract things from subordinate 3779 extensions ("I don't trust you to speak about net 7") [08:55:58] randy leaves the room [08:56:30] austein: I understand technical mechaism. the second-to-last bullet on slide 9.... "delte overlapping" is the magic. [08:57:32] randy: this is a whitewash over 7 TAs... could have put them in a directory... but you're reissuing.... you're not talking fixing little things, you're talking about fixing massive 3779 sections. kent: it's a matter of programming. [08:57:45] Doug_Otis leaves the room [08:59:56] hoffman: this slide is either wrong or you've made an assumption about what RPs what to manage that is incorrect. [09:00:09] kent: disagree. been told that RPs want to override IANA's allocations. [09:00:13] this mechanism allows that. [09:00:26] hoffman: as compared to putting IANA in the same slide..... [09:00:40] hoffman: parties won't know about unallocated addrs. [09:00:49] ggm is now known as ggm [09:00:49] ggm is now known as not-ggm [09:00:49] not-ggm is now known as ggm [09:00:53] ggm leaves the room [09:01:04] excellent entertainment value [09:01:12] randy: unless iana issues a ROA for unallocated, which I suggest they don't.... [09:01:33] haris.shamsi joins the room [09:01:52] [not as scribe: having gone back and listened to SF sidr recording, I think there was similarly harsh stuff said at that mtg] [09:05:50] not-ggm joins the room [09:05:50] not-ggm is now known as ggm [09:05:50] ggm is now known as ggm [09:06:03] ggm leaves the room [09:06:47] lebobits leaves the room: Logging off [09:07:04] who is at the microphone? [09:07:11] vijay.gill leaves the room [09:07:44] some guy in a Cisco shirt ;) [09:08:22] more specific please? [09:08:42] randy: steve is merely presenting a hack that an be used w/ no changes to current stuff. [09:09:21] randy and steve agreeing? Wow, now we know we've slipped into another dimension! [09:09:49] local. policy. happens. [09:10:08] please - could you ask folk to observe microphone etiquette please [09:10:23] Its impossible to follow this without some clarity of who is speaking [09:10:25] i'll try [09:10:44] thank you [09:10:45] randy is speaking [09:10:51] ggm joins the room [09:10:51] ggm is now known as not-ggm [09:10:51] not-ggm is now known as ggm [09:10:52] sam standing [09:10:53] (Sorry, missed, it was Wes George at the mic.) [09:10:59] I'll scribe momentarily for him [09:11:07] rob at mic [09:11:16] sam at mic [09:11:29] steve back in control [09:11:36] steve continues apace. [09:11:36] (well....) [09:11:48] thank you for that reminder to the room Sam - much appreciated [09:15:31] [not as scribe, tongue-in-cheek: does the CP apply to RP's doing this nonsense?] [09:15:52] ggm is now known as not-ggm [09:15:52] not-ggm is now known as ggm [09:15:52] ggm is now known as ggm [09:16:13] that depend on CP's, and this is not the nonsense [09:16:16] ASHIDA joins the room [09:16:42] Randy. [09:16:44] The CP doesn't put limits on the behaivor of Relying Parties ... The CA's sign on to the CP when they issue a cert with the OID for the CP, but the relying parties never sign on to the CP [09:17:19] this has all the problems of a unique DNS root. but the DNS root is that you and I could have different semantics for the same syntax. in this case, you and I can have different iew of the rtg world. I don't think I want to debug that. [09:17:45] i see the very sparing use of this hack to handle 1918, but I worry about us editting piblic space so that rtg across the net has an inconsistent view. [09:17:59] ggm leaves the room [09:18:03] kent; would only happen if inconsistentcy between rpki and what you think you know better [09:18:23] randy: and all the rirs think they know better than iana, so the problem stops there. any use beyond 1918 is dangerous. [09:18:28] kent; are defense uses valid: [09:18:29] phew, we're back to the normal dimension where randy and steven disagree [09:18:35] randy: no, they're not on the net [09:18:47] sandy: is there a way to prevent the stuff steve describes [09:18:49] randy: no [09:18:52] Hari leaves the room [09:19:29] randy: this complexity in the real world to solve political problem is bad. [09:19:35] sandy: but no way to prevent, right? [09:20:08] Rob A: important to distinguish between 2 cases: when RP thinks it knows more than what is published out there. and when it knows less [09:20:36] ex of 1: in DoD, when Army trusts DoD more than IANA. or 1918 space. [09:20:52] ggm joins the room [09:20:52] ggm is now known as not-ggm [09:20:52] not-ggm is now known as ggm [09:20:59] ASHIDA leaves the room: Replaced by new connection [09:21:03] what concerns me re: multiple root thing: is when RP has not enough info to decide bwtween roots [09:21:03] ASHIDA joins the room [09:21:59] i'm scribe now [09:22:20] rob: only want one root in public addr space [09:22:33] rob: not sure this hacks works at all levels [09:22:48] depends on 3779 (?) path validation check [09:23:20] each cert must be covered by its parent? or must go all the way up the chain? need to work it through and see [09:23:20] jpc leaves the room [09:23:30] steven: it works (missed justification) [09:23:51] kurtis leaves the room [09:24:03] Rob: this does not solve the overwhelming problem we have with the policy issues [09:24:07] not a magic bullet [09:24:10] weiler leaves the room [09:24:25] Terry: how does this work when the relying party can't create the signed objects in RSA? [09:24:39] must they use some other algo? [09:25:19] Steven: that nation will still be able to use their own algo of choice. If they say you can't use anything but their algo to verify, then u r up a creek [09:25:33] technically, in a pki it's possible to change algos as you go along [09:25:40] weiler joins the room [09:25:43] sean.s.shen leaves the room [09:25:51] if chose an algo that the rest of the world doesn't use, that's not playing well w/ others [09:25:53] ggm is now known as not-ggm [09:25:53] not-ggm is now known as ggm [09:25:53] ggm is now known as ggm [09:26:04] Sandy: we haven't dealt with algo diversity in the RPKI yet [09:26:25] Danny McPherson: if I was ISP and they decided to do this hack, then I would configure them as trust anchor? [09:26:29] SK: no [09:26:44] SK: you configure in your policy that they are holder of that addr space [09:27:01] SK: if you want to make life easy, you could delegate that trust entirely to them, but you don't have to [09:27:10] DM: trading autonomy for security [09:27:59] DM: if have multiple roots, then it prolongs ITU-T's influence in Internet [09:28:10] Andre R: missed question [09:28:18] thanks for taking over. my net's still unreliable, but I'll try.... [09:28:40] AR: if you are just relying party, you have local control and you can achieve the same thing with declared BGP policy w/o needing the whole RPKI system [09:28:51] ack [09:28:53] I'm out [09:28:55] phew [09:29:06] kent: you're not rebuilding an RPKI, juct changing the things you want [09:29:23] AR: are there cases you want to go bottom-up? [09:29:29] now I can get back to snide remarks [09:29:53] kent: we do discovery bottom-up but validation top-down.... by reissuing certs....we change how validation has to happen. [09:30:32] in the rpki, we allow people to find things, in discovery, with SIA extensions. [09:30:45] once it's found, it's when you import that you get to override for validation. [09:30:51] haris.shamsi leaves the room [09:30:52] they're different. [09:30:53] ggm is now known as ggm [09:30:53] ggm is now known as not-ggm [09:30:53] not-ggm is now known as ggm [09:31:07] out of time. [09:31:08] David Cooper leaves the room [09:31:08] mlepinski leaves the room [09:31:14] jpc joins the room [09:31:19] Michael Behringer leaves the room [09:31:30] bye bye folks [09:31:44] thanks Sam and Greg for the jabber notes [09:31:47] weiler leaves the room [09:31:48] pssara leaves the room [09:31:50] Brenden Kuerbis leaves the room [09:31:53] roque leaves the room [09:31:58] Merridew leaves the room: QIP Infium: Спокойное общение [09:31:59] rhe leaves the room [09:32:00] jpc leaves the room [09:32:04] Robert Kisteleki leaves the room [09:32:06] ack [09:32:08] Jan Boogman leaves the room [09:32:13] gih leaves the room [09:32:13] lellel leaves the room [09:32:13] lebo leaves the room [09:32:20] tachibana@jabber.org leaves the room [09:32:23] aalain leaves the room [09:32:36] yoshihiko.saeki@jabber.org leaves the room [09:32:54] ggm leaves the room [09:33:04] shinmiyakawa leaves the room [09:33:59] dougm.home leaves the room [09:35:06] Robert Kisteleki joins the room [09:35:59] psavola leaves the room [09:36:48] Robert Kisteleki leaves the room [09:44:03] ASHIDA leaves the room: Replaced by new connection [09:49:03] Hari joins the room [09:50:21] Hari leaves the room [09:50:59] Doug_Otis joins the room [09:57:19] Doug_Otis leaves the room [10:01:24] Brenden Kuerbis joins the room [10:01:45] Robert Kisteleki joins the room [10:02:54] rababy leaves the room: Computer went to sleep [10:03:42] Robert Kisteleki leaves the room [10:11:54] Brenden Kuerbis leaves the room [10:11:55] Jan Boogman joins the room [10:12:11] Jan Boogman leaves the room [10:15:38] Linus Nordberg joins the room [10:25:04] Linus Nordberg leaves the room [10:58:25] Aleksi Suhonen leaves the room [10:58:26] Linus Nordberg joins the room [11:17:01] Brenden Kuerbis joins the room [11:39:15] not-ggm joins the room [11:41:42] not-ggm leaves the room [12:01:40] Linus Nordberg leaves the room [12:15:04] Brenden Kuerbis leaves the room [14:18:17] Linus Nordberg joins the room [15:46:06] Linus Nordberg leaves the room