IETF
tsvarea
tsvarea@jabber.ietf.org
Thursday, 28 July 2011< ^ >
Room Configuration

GMT+0
[12:44:20] wesley.m.eddy joins the room
[12:44:50] wesley.m.eddy has set the subject to: IETF81
[12:52:01] jlcJohn joins the room
[12:59:59] =JeffH joins the room
[13:00:01] mcharlesr joins the room
[13:00:16] <mcharlesr> is anyone remote?
[13:00:28] mcharlesr has access to mic.
[13:00:42] <jlcJohn> me
[13:01:22] wolfgang.beck01 joins the room
[13:02:36] <wesley.m.eddy> can you hear the audio? we are starting
[13:02:43] Andrew McGregor joins the room
[13:08:37] <jlcJohn> The audio is good> :^)
[13:19:00] <wesley.m.eddy> david bryan is speaking now
[13:19:56] <=JeffH> wes' slides are not in the meeting materials repository ??
[13:20:25] <=JeffH> ??
[13:20:58] <wesley.m.eddy> they were posted yesterday: http://www.ietf.org/proceedings/81/slides/tsvarea-5.pptx
[13:21:16] <wesley.m.eddy> David's slides being presented now are: http://www.ietf.org/proceedings/81/slides/tsvarea-2.pptx
[13:21:28] <=JeffH> yes, wrt David's, got those
[13:21:40] <=JeffH> thx
[13:22:32] <jlcJohn> I'm unclear what slide number...
[13:22:39] <wesley.m.eddy> he's on 5
[13:22:46] <jlcJohn> thx
[13:22:58] <wesley.m.eddy> title is "Video Streaming/P2PTV"
[13:23:16] <wesley.m.eddy> now 6
[13:23:23] <=JeffH> k, thx, .... ah ha, I hadn't reloaded my cached copy of the meeting materials page since yesterday, sorry
[13:23:58] <wesley.m.eddy> now 7
[13:24:43] <wesley.m.eddy> now 8
[13:25:06] <wesley.m.eddy> now 9
[13:25:58] <wesley.m.eddy> now 10
[13:26:49] <wesley.m.eddy> now 11
[13:27:47] <wesley.m.eddy> now 12
[13:29:29] <wesley.m.eddy> now 13
[13:29:31] mcharlesr leaves the room
[13:30:00] <wesley.m.eddy> now 14
[13:30:06] mcharlesr joins the room
[13:30:22] <wesley.m.eddy> now 15
[13:31:15] <wesley.m.eddy> now 16
[13:31:24] <=JeffH> fyi wrt reputation....
[13:31:27] <=JeffH> David - P2P Applications and the IETF What do we have? : What are we missing?
[13:31:38] <=JeffH> oops pls ignore prior
[13:31:45] wolfgang.beck01 leaves the room
[13:31:46] <=JeffH> draft-kucherawy-reputation-media-type-00
draft-kucherawy-reputation-model-00
draft-kucherawy-reputation-query-dns-00
draft-kucherawy-reputation-query-http-00
draft-kucherawy-reputation-query-udp-00
draft-kucherawy-reputation-vocab-email-00
draft-kucherawy-reputation-vocab-identity-00
[13:32:45] <wesley.m.eddy> Rich Woundy now, slides are: http://www.ietf.org/proceedings/81/slides/tsvarea-3.pptx
[13:33:36] <wesley.m.eddy> now "Presentation Overview" slide
[13:34:32] <wesley.m.eddy> picture of traffic jam
[13:36:19] <wesley.m.eddy> now "Composition of Network Traffic" slide
[13:37:30] <wesley.m.eddy> Stewart Cheshire: columns sum to 100, not rows
[13:38:05] <wesley.m.eddy> jim gettys asked about ACKs
[13:38:43] <wesley.m.eddy> jim notes upstream ACKs become bottleneck for downstream performane
[13:38:54] mph joins the room
[13:39:11] <wesley.m.eddy> on "Evolution in Aggregate Traffic" slide
[13:41:10] <wesley.m.eddy> michael richardson notes it would be good to have absolute numbers as well as the percentages
[13:42:18] <wesley.m.eddy> now on "P2P Traffic Impacts" slide
[13:44:37] narten joins the room
[13:45:11] <wesley.m.eddy> on "Simplified Cable Access Network"
[13:45:22] <wesley.m.eddy> on "Impact of the P2P Overlay"
[13:46:08] <wesley.m.eddy> on "TSV WG Efforts ..."
[13:46:17] mph leaves the room
[13:47:04] <wesley.m.eddy> on Benefits of Swarm Localization
[13:48:17] <wesley.m.eddy> on "P2P Overlay with localization"
[13:48:38] narten leaves the room
[13:49:03] <wesley.m.eddy> on "Benefits of P2P Caching"
[13:49:56] <wesley.m.eddy> question: is it legal to do content caching?
[13:50:26] <wesley.m.eddy> on "P2P Overlay Using Network Caching"
[13:50:28] <Andrew McGregor> Who was that?
[13:50:35] <wesley.m.eddy> unknown
[13:51:33] <wesley.m.eddy> on "Economics behind network caching"
[13:52:08] shikob joins the room
[13:53:41] <=JeffH> what's a "CMTS" box ?
[13:53:49] <wesley.m.eddy> cable infrastructure
[13:54:08] <wesley.m.eddy> http://en.wikipedia.org/wiki/Cable_modem_termination_system
[13:54:18] <=JeffH> super thx
[13:54:24] <wesley.m.eddy> on Rich's last slide "TSV P2P WG Efforts"
[13:54:51] <wesley.m.eddy> bob briscoe
[13:56:53] <wesley.m.eddy> bob points out ledbat may be more helpful in the home than the cmts
[13:57:24] <wesley.m.eddy> cheshire - broad comments
[13:58:23] <wesley.m.eddy> pointing out that some of these things aren't just P2P
[14:00:06] <wesley.m.eddy> moobeen pashnavich (?) - asks how old the numbers are
[14:02:00] <=JeffH> specifically the "economics behind network caching" numbers
[14:03:41] <wesley.m.eddy> al morton
[14:04:18] <wesley.m.eddy> layer 8 problem
[14:04:37] <wesley.m.eddy> you have to pay to get burned in the end
[14:06:02] <wesley.m.eddy> moving to jana: http://www.ietf.org/proceedings/81/slides/tsvarea-0.pdf
[14:06:51] <wesley.m.eddy> on "... but the Internet remains loyal"
[14:07:16] <wesley.m.eddy> on "How deep .."
[14:08:05] <wesley.m.eddy> on taxonomhy slide
[14:08:19] mph joins the room
[14:09:05] <wesley.m.eddy> (slide is animated, but you probably can't see that in the pdf)
[14:09:25] <jlcJohn> :^)
[14:09:54] <wesley.m.eddy> on "why does this taxonomy matter"
[14:10:16] <jlcJohn> BTW, I'm _really_ glad I don't have to watch NeoOffice mangle a PPT
[14:10:59] <wesley.m.eddy> on "deployment impossibility"
[14:11:31] <wesley.m.eddy> on "what have we done so far?"
[14:11:54] <wesley.m.eddy> (also animated)
[14:12:18] <wesley.m.eddy> on "the final stage"
[14:12:24] <=JeffH> <applause on slide 8>
[14:13:15] <wesley.m.eddy> on "the minion suite"
[14:14:01] <wesley.m.eddy> on "outline"
[14:14:27] <wesley.m.eddy> on "what's in the minion suite" (animation again)
[14:15:06] <wesley.m.eddy> on "TCP Minion" slide
[14:15:43] <wesley.m.eddy> on "delivery in standard tcp"
[14:16:26] <wesley.m.eddy> on "delivery in tcp minion"
[14:16:51] <wesley.m.eddy> on "problem: network resegmentation"
[14:17:35] <wesley.m.eddy> (also animated)
[14:17:45] <wesley.m.eddy> on "COBS encoding"
[14:18:25] mph leaves the room
[14:18:34] <wesley.m.eddy> on "App messages ..."
[14:19:25] <wesley.m.eddy> on "app-observed delay distribution"
[14:20:11] <wesley.m.eddy> on "voip: perceptual quality impact"
[14:20:12] narten joins the room
[14:20:48] <wesley.m.eddy> on "tcp minion: what's next"
[14:21:57] <wesley.m.eddy> on "app with message priorities"
[14:23:33] Bill Check joins the room
[14:23:43] <wesley.m.eddy> ekr question - transparent proxies
[14:23:49] <wesley.m.eddy> "this doesn't work"
[14:24:29] <wesley.m.eddy> stuart - clarifies point
[14:24:37] jgunn joins the room
[14:26:56] <wesley.m.eddy> on "tls minion (summary)"
[14:27:44] <wesley.m.eddy> on "our implementation of the minions"
[14:28:04] <wesley.m.eddy> on "in conclusion"
[14:28:20] <jlcJohn> channel if necessary: where do we discuss this?
[14:28:28] <wesley.m.eddy> matthew kaufman
[14:28:53] <wesley.m.eddy> tsv-area@ietf.org please
[14:29:37] <wesley.m.eddy> on tcpcrypt presentation onw
[14:35:33] <wesley.m.eddy> on "after initial handshake" slide
[14:36:13] <wesley.m.eddy> on "how to check the session id"
[14:36:33] <wesley.m.eddy> on "authentication example"
[14:37:26] <wesley.m.eddy> ekr at mic
[14:37:51] <wesley.m.eddy> people using https have 4 mechanisms to do this and don't use any
[14:38:10] <wesley.m.eddy> mark's answer notes the user interface
[14:38:43] <wesley.m.eddy> matthew kaufman asks about lack of perfect fwd secrecy
[14:39:13] <wesley.m.eddy> mark clarifies that it's a local decision how often to generage ephemeral pub key
[14:39:41] <wesley.m.eddy> tim shepard (not at mic) says that's the same as ssh
[14:40:01] <wesley.m.eddy> on example "signing a batch of session IDs"
[14:40:26] <wesley.m.eddy> (animation which isn't in PDF file on web)
[14:41:35] mcharlesr leaves the room
[14:41:38] <wesley.m.eddy> on "tcpcrypt in detail"
[14:41:50] <wesley.m.eddy> on "outline of tcpcrypt key exchange"
[14:43:12] EKR joins the room
[14:43:18] <EKR> http://www.imperialviolet.org/2010/06/25/overclocking-ssl.html
[14:43:24] <wesley.m.eddy> michael richardson - note that crypt-hello in syn is not necessary
[14:44:18] <wesley.m.eddy> mark refers to michio's study on middleboxes in the wild
[14:45:30] <wesley.m.eddy> on "key scheduling"
[14:45:48] <wesley.m.eddy> on tcpcrypt in tcp packets
[14:46:00] mcharlesr joins the room
[14:46:06] <EKR> basically, this whole notion that SSL/TLS is too expensive is 5 years out of date.
[14:46:29] <EKR> Most important paragraph: "n January this year (2010), Gmail switched to using HTTPS for everything by default. Previously it had been introduced as an option, but now all of our users use HTTPS to secure their email between their browsers and Google, all the time. In order to do this we had to deploy no additional machines and no special hardware. On our production frontend machines, SSL/TLS accounts for less than 1% of the CPU load, less than 10KB of memory per connection and less than 2% of network overhead. Many people believe that SSL takes a lot of CPU time and we hope the above numbers (public for the first time) will help to dispel that."
[14:47:58] <wesley.m.eddy> (ekr: you may want to use the tsv-area@ietf.org mailing list)
[14:48:06] Bill Check leaves the room
[14:48:43] <wesley.m.eddy> bob briscoe responding to david's point that the server supporting it will make sure that the firewall near it does
[14:48:59] <wesley.m.eddy> on "crypto state can be cached"
[14:49:43] <wesley.m.eddy> on "tcpcrypt implementations"
[14:50:44] <wesley.m.eddy> on "apache"
[14:51:34] <wesley.m.eddy> michael richardson - replace tcp-md5 for bgp
[14:51:48] Hayim joins the room
[14:52:01] <wesley.m.eddy> mark - this is over tcp not in-tcp
[14:52:01] <wesley.m.eddy> different scope
[14:52:10] <wesley.m.eddy> richard barnes - we already have tcp ao
[14:52:10] narten leaves the room
[14:53:22] <wesley.m.eddy> on "why tcpcrypt"
[14:53:25] EKR leaves the room
[14:54:22] <wesley.m.eddy> on 2nd "why tcpcrypt"
[14:55:08] narten joins the room
[14:56:42] narten leaves the room
[14:57:06] <wesley.m.eddy> on summary slide, going to brief demo
[14:57:33] <wesley.m.eddy> demo through tcpcrypt.org website and hall of fame there
[14:57:49] <wesley.m.eddy> david mcgrew
[14:58:18] <wesley.m.eddy> concern about inserting in tcp layer and in claimed advantages
[14:58:54] <wesley.m.eddy> mentions claims about rsa don't mention ecc
[15:00:36] <wesley.m.eddy> (refer back to why tcpcrypt slide)
[15:01:20] <wesley.m.eddy> mark mentions that this has to be "under the socket"
[15:02:19] <wesley.m.eddy> matthew kaufman - how do you protect against downgrade attack race
[15:03:07] <wesley.m.eddy> if you tease apart the pieces, can you sit outside the company and downgrade connections
[15:03:38] <wesley.m.eddy> ekr - based on particular use model (server with lots of connections)
[15:03:51] <wesley.m.eddy> mark - latency and conn/sec are different metrics
[15:04:31] <wesley.m.eddy> ekr - 1 rtt for tls is possible rather than 2
[15:04:45] <wesley.m.eddy> (real concern in 2006, not current)
[15:05:22] <wesley.m.eddy> now we get to the relevant point ... charges for addresses and certs
[15:05:22] Lars joins the room
[15:06:09] <wesley.m.eddy> mark and ekr disagree; ekr says it's been measured
[15:06:21] <wesley.m.eddy> richard barnes: bump in the wire is a use case, problem is that it works in reverse direction (transparently removable by a bump)
[15:07:54] <wesley.m.eddy> richard: mentions btns and other ways of doing opportunistic crypto
[15:08:08] <wesley.m.eddy> mark: rollout is the fundamental issue
[15:08:13] <wesley.m.eddy> jeff hodges (paypal) likes use of opportunistic crypto
[15:09:18] <wesley.m.eddy> points out that there are other issues with deployment (mashed up web apps), mentions adam langley's use of opportunistic crypto (different approach to saving an RTT)
[15:10:08] <wesley.m.eddy> mcgrew - need to worry about active in addition to passive eavesdropping
[15:10:17] <wesley.m.eddy> firesheep++ does matthew's downgrade
[15:10:36] <wesley.m.eddy> switching to Yingjie presentation
[15:11:09] <wesley.m.eddy> http://www.ietf.org/proceedings/81/slides/tsvarea-1.pdf
[15:11:24] <wesley.m.eddy> on background introduction
[15:11:51] <wesley.m.eddy> on in-site and between-sites migration
[15:12:40] <wesley.m.eddy> on "operational state migration"
[15:14:33] <wesley.m.eddy> on "two categories of operational state"
[15:15:51] <wesley.m.eddy> on "example architecture"
[15:16:09] <wesley.m.eddy> on "example flow"
[15:17:44] narten joins the room
[15:18:34] <wesley.m.eddy> on "problem decomposition"
[15:19:31] <wesley.m.eddy> questions
[15:19:50] <Andrew McGregor> Someone channel me please: isn't FORCES relevant here?
[15:20:14] <wesley.m.eddy> briscoe - going back to time-sensitive, doesn't know if it affects protocol, more like how things are coded
[15:21:21] shep joins the room
[15:21:26] gorryf joins the room
[15:21:45] <wesley.m.eddy> david black - clarifies his remarks (not company)
[15:21:52] <wesley.m.eddy> question on time flow slide
[15:22:22] <wesley.m.eddy> white spot is time-criticial and done when vm is not running, must be minimized, head for doing capability check beforehand
[15:22:40] <wesley.m.eddy> vm migration already does this
[15:24:56] <wesley.m.eddy> david - thinks its necessary to work with what's out there instead of turning vm manager into 2-phase commit protocol
[15:24:58] <wesley.m.eddy> moving to mirja
[15:25:38] <wesley.m.eddy> on motivation
[15:26:56] <wesley.m.eddy> on accurate ecn feedback in tcp
[15:27:13] shep leaves the room
[15:27:44] <wesley.m.eddy> meeting adjourned
[15:28:14] gorryf leaves the room
[15:28:37] narten leaves the room
[15:28:54] Hayim leaves the room: Computer went to sleep
[15:29:17] Andrew McGregor leaves the room
[15:29:34] wesley.m.eddy leaves the room
[15:33:56] Lars leaves the room
[15:43:16] mcharlesr leaves the room
[15:43:45] =JeffH leaves the room
[15:46:17] shikob leaves the room
[15:50:34] jgunn leaves the room
[15:59:27] wesley.m.eddy joins the room
[16:39:57] wesley.m.eddy leaves the room
[16:59:58] Lars joins the room
[17:02:48] Andrew McGregor joins the room
[17:04:30] narten joins the room
[17:06:50] EKR joins the room
[17:10:58] Andrew McGregor leaves the room
[17:32:13] EKR leaves the room
[17:34:19] EKR joins the room
[17:45:14] EKR leaves the room
[17:46:20] EKR joins the room
[17:47:33] EKR leaves the room
[17:48:43] Lars leaves the room
[17:50:22] EKR joins the room
[17:55:43] EKR leaves the room
[17:58:04] EKR joins the room
[18:28:59] EKR leaves the room
[18:31:35] jlcJohn leaves the room
[18:32:55] EKR joins the room
[18:33:53] narten leaves the room
[18:49:50] narten joins the room
[18:52:01] narten leaves the room
[19:03:56] EKR leaves the room
[19:11:16] EKR joins the room
[19:12:45] EKR leaves the room
[19:15:19] EKR joins the room
[19:15:34] EKR leaves the room
[19:20:46] EKR joins the room
[19:21:52] EKR leaves the room
[19:25:52] narten joins the room
[19:29:49] narten leaves the room
[19:53:37] EKR joins the room
[19:56:59] EKR leaves the room
[20:01:19] EKR joins the room
[20:05:28] EKR leaves the room
[20:05:29] EKR joins the room
[20:09:28] EKR leaves the room
[20:15:38] EKR joins the room
[20:17:20] EKR leaves the room
[20:21:16] jlcJohn joins the room
[20:31:12] EKR joins the room
[20:33:34] EKR leaves the room
[20:37:45] EKR joins the room
[20:44:55] EKR leaves the room
[20:48:59] EKR joins the room
[20:55:15] EKR leaves the room
[20:55:30] EKR joins the room
[21:08:38] EKR leaves the room
[21:37:03] EKR joins the room
[22:51:02] EKR leaves the room
[22:51:56] EKR joins the room
[23:44:31] EKR leaves the room
[23:49:42] EKR joins the room
Powered by ejabberd Powered by Erlang Valid XHTML 1.0 Transitional Valid CSS!