14:59 < Chris_RIPENCC> Hi all, the afternoon plenary session is about to begin - I'll be your friendly chat monitor! Any questions or comments, just post them here and I can read them out over the microphone.
15:00 < Chris_RIPENCC> First up is a lightning talk: Challenges in Endpoint DNSSEC by OndΕ™ej Caletka, CESNET
15:01 < ripe963> nice
15:05 -!- keith_nm is now known as keith_24865
15:08 < mickod-2110-mbp> DOH!
15:09 < Chris_RIPENCC> Are there any questions for OndΕ™ej?
15:10 < Chris_RIPENCC> Next up is How the Hell Should We Fund Open Source? by Jeff Osborn, ISC
15:14 < Chris_RIPENCC> Any questions or comments for Jeff?
15:17 < ripe742> In reply to that mic: BSD license consumers give a LOT back these days
15:18 < Chris_RIPENCC> would you like me to read that out?
15:18 < ripe742> No need; it's recorded here
15:19 < Chris_RIPENCC> it is
15:21 < shane> "Sales and marketing are not in the talent set of most open source projects"
15:21 < shane> That's an understatment!!!
15:22 < Chris_RIPENCC> Next up is Protecting the Routing Table: Golden Prefixes by Job Snijders, NTT
15:23 < tm-exa> The money is often made by the added value reseller who are taking the software and package it in nice easy to deploy appliance. As long as they can do this without participating to the cost, then it will not really change and the money will flow to them. The licence is therefore indeed very important and why in some case AGPL + proprietary licence may be a good licence as these player will want to keep their "added value sauce" proprietary
15:23 < tm-exa> and therefore force them to contribute by paying ...
15:25 < Micromus> yeah, dual licensing is a good way to go I think
15:25 < Micromus> makes it possible for other volunteurs to contribute and create hobby projects, but prevents explotation by those who actually make the money of the products
15:31 < ripe968> Sandy Murphy: How do you deal with new announcements? who is the auditor? "only usable local policy is to drop invalids" - not true. i know of US research net that is using up-pref for valid, down-pref for invalid. your trust model is awfully trusting.
15:32 < Azi-47720> down-pref-on-invalid does not protect against hijacked more-specifics
15:33 < ripe968> Sandy Murphy: yes. as first deployment, it works.
15:33 < Chris_RIPENCC> i'll read your comment sandy
15:35 < Chris_RIPENCC> That's all for the lightning talks - next up is Stop Thinking IPv4; IPv6 is Here by Jen Linkova, Google
15:35 < ripe968> Sandy Murphy: fixing IRR outside RIPE is very hard - RADB has no access to info as to authenticating person registering as person authorized for resoruce.
15:35 < job> Sandy Murphy, please look for Jared at IETF next week
15:35 < ripe968> Sandy Murphy mauch?
15:35 < job> yes
15:36 < ripe968> ok.
15:36 < job> he and i are circling the idea and assessing traction and usability
15:36 < job> thank you for your comments, would love to chat with an icecream face2face but i suspect you are not here in london
15:36 < shane> In 4 years we will have 143% IPv6!!! πŸ˜€
15:36 < ripe968> Sandy Murphy: sadly, regrettably, no. ice cream sounds nice, she said plaintively.
15:37 < rhe-786> Dunno shane, that graph looked as though it was starting to flatten off.
15:37 < shane> Isn't there supposed to be something to the west of Belgium? πŸ˜‰
15:38 < lochii> job:
15:38 < job> ha nice
15:38 < lochii> rocking the foundry configs
15:38 < lochii> the domain is now being squatted
15:39 < lochii> but for some time, joao deconfigured the vhost and you instead got pictures of him and his family
15:39 < brian-1213> shane: Couple of things...
15:39 < shane> I love link local addresses.
15:41 < sps4-ripe> shane: I think she does as well - Considering her last name πŸ˜‰
15:42 < shane> sps4-ripe: Is that worth a groan? Okay. *groan*
15:42 < sps4-ripe> shane: that's ok πŸ™‚ I groaned myself at it
15:42 < Habbie> totally groanworthy
15:42 < lochii> I think it is possible in ipv4, in two ways, first of which is using ipv4 link local (169.254.x.x), second of which is with unnumbered p2p (old hdlc/frame/ppp or now unnumbered ethernet)
15:42 < Habbie> shane, do you still carry DNS IS SEXY-stickers?
15:43 < shane> Habbie: Yes, they're in my jacket right now but I'll have them on me when I go to the social tonight. If you want one earlier grab me after this session.
15:44 < Habbie> shane, ok!
15:44 < Habbie> either works for me
15:46 < tm-exa> As here may be the place ... Can someone contact me off channel if they can provide me a pcap of a BGP message with an ASPATH with confederation in IPv4 and IPv6
15:46 < shane> I wish Google didn't hate DHCPv6 so much. πŸ™
15:47 < shane> Although that slide was fair.
15:50 < lochii> I think there isn't wonderful RDNSS adoption right now
15:51 < brian-1213> I wonder how many servers/boxes around the internet are called Wintermute.
15:51 < brian-1213> And are there enough of them to gain sentience.
15:51 < Azi-47720> been a long while since I came across a winternmute
15:56 < Micromus> for ISPs which are by law held accountable for end users, is there any alternative to dhcpv6?
15:57 < Chris_RIPENCC> Micromus, did you want that question asked over the mic? or just asking in the chat room?
15:59 < Chris_RIPENCC> any questions for Jen?
16:00 < shane> @Micromus: static /48 allocation?
16:02 < Micromus> Chris_RIPENCC: no, I'm just throwing it out here πŸ™‚
16:02 < Micromus> shane: but that would be much more work than dhcpv6, no?
16:03 < Micromus> Not all IPv4 ways of thinking are right, but not all of them are wrong either, as in the case of dhcp, in many cases that IS the right answer IMHO
16:03 < lochii> Micromus: good point about logging
16:03 < shane> Micromus: Depends on the environment... having written ISC's old DHCPv6 server I tend to think it's the most straightforward solution in such an environment, but I do realize that people have other ideas. πŸ™‚
16:04 < wiro> but that's what she said, if you have good reason to use dhcpv6, you can
16:04 < wiro> but think first if you need it
16:05 < wiro> I think this guy had very good point. In our company ipv6 is not technical problem
16:05 < rhe-786> People still see IPv6 as "innovative and forward looking?" πŸ™
16:05 < wiro> just management don't see added value
16:05 < Chris_RIPENCC> Next is SIIT-DC: IPv4 Service Continuity for IPv6 Data Centres by Tore Anderson, Redpill Linpro
16:08 < chrisr-31170> speaking with one of the carriers yesterday the biggest headache they've had is being able to cope with infrastructure for logging with respect to court orders, etc
16:09 < lochii> where is the ipv6 hipster cat?
16:16 < Micromus> chrisr-31170 +1
16:16 < shane> Days since IPv6 to IPv4 mapping technology: 0
16:17 < Micromus> hehe, but I like this, no state, no need for support end-to-end, just a stateless box in the middle doing 1:1 mapping
16:18 < shane> Yes, it is not too bad.
16:19 < Micromus> Compared to other things which needs either expensive and hard-to-scale hardware, or deploying stuff on loads of different gear (which will never happen)
16:20 < wiro> I have one question, do you know about some implementation of GEOIP for translated prefix ? If your application or webserver must act on that information.
16:20 < ripe491> Is it possible to do such mappings based on dest. port?
16:20 < Micromus> Many GEOIP stuffs have APIs? Should be implementable in a day
16:20 -!- ripe491 is now known as samm
16:20 < murb> wiro: teach your geoip service how to convert mapped addresses?
16:21 < Micromus> or just create a middle-api which strips of the first 96 bits
16:21 < murb> Micromus: exactly.
16:21 < Micromus> samm: no, I don't think that is in the scope
16:22 < Micromus> But would be easy to add by adding port to the translated ipv6 addrs
16:22 < Micromus> Might create problems that the translator-box needs to keep state, which defeats the whole purpose
16:22 < wiro> Do you have some experience doiging this in hi traffic services ?many milions of users in same time
16:23 < wiro> *doing πŸ™‚
16:23 -!- Otacon22 is now known as daniel
16:23 < Micromus> As he just said, should be possible to do 10GE wirespeed on a regular linux server
16:24 < samm> Micromus thanks, i see.
16:24 < Micromus>
16:25 < Micromus> Example of a framework to create a middlebox that I assume can modify/translate packets reeeally efficiently
16:28 < Chris_RIPENCC> Are there any questions for Tore?
16:50 < mickod-2210-mbp> nick mickod-2110
16:50 -!- mickod-2210-mbp is now known as mickod-2110
17:00 -!- chrisr is now known as kit_chrisr