IETF
dtn@jabber.ietf.org
Thursday, November 13, 2014< ^ >
Room Configuration
Room Occupants

GMT+0
[01:05:56] Meetecho joins the room
[01:12:09] Meetecho leaves the room
[01:15:07] Keith Scott joins the room
[01:15:07] Will Ivancic joins the room
[01:15:14] Meetecho feed joins the room
[01:16:23] Brian Haberman joins the room
[01:16:51] marc.blanchet.qc joins the room
[01:16:59] <Will Ivancic> Signed in.  Camera and sound appears to be working.
[01:17:16] Alessandro Amirante joins the room
[01:18:06] Scott Burleigh joins the room
[01:18:49] John Leslie joins the room
[01:19:11] Keith Scott leaves the room
[01:19:29] jlcJohn joins the room
[01:19:38] Keith Scott joins the room
[01:20:00] Scott Burleigh leaves the room
[01:20:11] Alessandro Amirante leaves the room
[01:20:34] Alex Amirante joins the room
[01:22:27] Scott Burleigh joins the room
[01:23:16] Elwyn Davies joins the room
[01:23:23] Keith Scott leaves the room
[01:24:04] <Will Ivancic> I see Scott.
[01:24:20] Keith Scott joins the room
[01:24:29] <Alex Amirante> Yep, mee too :)
[01:24:48] <Will Ivancic> Now I hear sound
[01:24:56] <Scott Burleigh> And I see the head table.
[01:24:59] Kurt Andersen joins the room
[01:25:00] Fuyou Miao joins the room
[01:25:25] Elwyn Davies leaves the room
[01:25:47] <Will Ivancic> Yes, me too.  I have my camera off until I speak.
[01:26:07] elwynd joins the room
[01:27:26] Elwyn Davies joins the room
[01:27:37] Kurt Andersen leaves the room
[01:28:22] John Dowdell joins the room
[01:28:36] Elwyn Davies leaves the room
[01:28:45] <John Dowdell> Hi all, your jabber scribe here. Anyone here?
[01:29:04] <elwynd> yes
[01:29:20] <John Dowdell> Hi Elwyn
[01:29:27] <Alex Amirante> Presentation is over
[01:30:27] <John Dowdell> Meetecho just come up
[01:30:29] Keith Scott_5894 joins the room
[01:31:06] Keith Scott_5894 leaves the room
[01:31:09] Keith Scott leaves the room
[01:31:15] Joseph joins the room
[01:31:28] Anders Lindgren joins the room
[01:31:30] <John Dowdell> Scott Burleigh presenting topic on bundle protocol
[01:32:20] Elwyn Davies joins the room
[01:32:42] Keith Scott joins the room
[01:33:19] Henry Suthon joins the room
[01:34:30] Edward Birrane joins the room
[01:36:20] <elwynd> Extension blocks: Currently *too* mutable... no idea which ones might have been there
[01:37:13] <John Dowdell> At chair request, holding questions to the end, is that ok?
[01:37:33] <elwynd> sure.. more a commemnt than a question
[01:37:34] Suzanne Davidson joins the room
[01:38:46] mvidebaekpedersen joins the room
[01:39:44] mvidebaekpedersen leaves the room
[01:39:48] Henry Suthon_laptop joins the room
[01:40:59] Morten V. Pedersen joins the room
[01:40:59] Henry Suthon_laptop leaves the room
[01:42:09] Davud Eddy joins the room
[01:42:32] Franck Martin joins the room
[01:43:08] Kurt Andersen joins the room
[01:43:33] <Joseph> Having a flat address space would also be taxing for constrained devices.
[01:49:27] david eddy joins the room
[01:50:06] Keith Scott leaves the room
[01:50:11] Davud Eddy leaves the room
[01:51:51] jeferson.nobre joins the room
[01:54:22] <Edward Birrane> Question about checksums: If we mandate a checksum in the spec, how do we pick which checksum?
[01:54:42] tomek joins the room
[01:55:25] <Joseph> If you need something strong I would leave that to the application
[01:56:11] <Joseph> And use a more straightforward (fast) checksum for the protocol
[01:56:31] Morten V. Pedersen leaves the room
[01:57:01] <Joseph> That's my take anyways
[01:57:22] <Edward Birrane> Understood, but mandating a checksum for the primary block can't be left to the application, and in my experience checksums can be best suited to links with different characteristics?
[01:58:06] <Edward Birrane> It's a small matter, but I suspect it will lead to having two checksums, the default and something specific in another extension block anyway.
[01:59:07] <Brian Haberman> @Ed: Can we take the ICV model used by IPsec, where the checksum is variable length based on the algorithm used?
[01:59:15] <Elwyn Davies> Ext blocks.. at present the integrity systems in BSP don't cover the whole bundle so you have no idea whether you have a complete bundle
[02:00:07] <Elwyn Davies> e.g., LTP  red vs green portions
[02:00:08] <Joseph> To clarify, "By leaving it the application" I meant that if the application wanted a stronger checksum than the mandated one, then it can just add it in addition to the mandated one
[02:00:10] Alexa  Morris joins the room
[02:00:26] Alexa  Morris leaves the room
[02:00:44] Keith Scott joins the room
[02:01:00] <Will Ivancic> I have some talk of checksum in my discussion.  Perhaps we should wait until those slides are presented.
[02:01:14] <Joseph> I'm fine with that
[02:01:19] <Edward Birrane> I think SBSP provides block-level end-to-end checksums that address how to handle integrity failure at waypoints, but whether we want to roll out SBSP for any end-to-end integrity is another issue.
[02:01:30] <Edward Birrane> @will: fine with that too
[02:03:23] Meetecho joins the room
[02:06:17] <Elwyn Davies> Re ICN stuff: bundles may be created by nodes in the middle of the net that has a copy of a bundle that matches a request and is neither the publisher or the original sender of the bundle.
[02:07:06] <Edward Birrane> Comment: Technical items capture set that seems important and roughly in priority order.
[02:07:09] <Will Ivancic> Scott can you release the presentation
[02:07:23] <Alex Amirante> Scott can you please stop your presentation, so Will can load his slides?
[02:07:39] marc.blanchet.qc joins the room
[02:07:41] marc.blanchet.qc leaves the room
[02:07:59] <Will Ivancic> I cannot get my slides up until Scott releases
[02:08:22] <Will Ivancic> At least that appears to be the case.
[02:08:38] <Alex Amirante> Yes Will
[02:09:05] <Scott Burleigh> Presentation is over
[02:09:42] <Edward Birrane> We can't hear will.
[02:09:50] <John Dowdell> Will Ivancic now presenting
[02:10:00] <Will Ivancic> I am not muted.
[02:10:11] <Edward Birrane> Yes.
[02:10:12] <Elwyn Davies> dictionaries and full EIDs are two different issues.  The dictionaries are an implementation nightmare but we need EIDs (IMO)
[02:10:19] <Joseph> No audio from will
[02:10:38] <John Dowdell> Joseph do you have audio now?
[02:10:48] <Joseph> I can hear the room
[02:10:58] <Joseph> Now I can hear Will too
[02:11:03] <John Dowdell> can you hear Will?
[02:11:07] <John Dowdell> ok good
[02:12:16] <marc.blanchet.qc> Scott, if you are still there. I had a question, but given time, I did not ask. Here I ask in case you are still able to answer:
[02:12:28] Alexa  Morris joins the room
[02:12:36] <marc.blanchet.qc> you are proposing many stuff, more extension blocks, multicast, etc…
[02:13:08] <marc.blanchet.qc> I claim that most « core » use cases in DTN such as space, uav and others are « resource-constrained », embedded systems.
[02:13:11] Alexa  Morris leaves the room
[02:14:03] <marc.blanchet.qc> in embedded systems, (we have done a dtn implementation that runs on rtems…), the simplest, less options, etc… is the key .
[02:14:26] <marc.blanchet.qc> so should we have a goal to actually make BP and related protocols simpler rather than more complex?
[02:15:26] <Scott Burleigh> I agree that those use cases are resource-constrained, but these are features that reduce bandwidth consumption and/or procwssing load.
[02:15:56] <Scott Burleigh> Fewer options doesn't mean better for resource-constrained environments, I believe.
[02:17:19] <Scott Burleigh> That said, I completely agree that reducing complexity is very valuable.  That's exactly what numeric node IDs, for example, would do.  In ION, it would let me lose at least 800-1000 lines of code.
[02:17:30] <Elwyn Davies> The problem of extension blocks was orthogonal to the difficulties of integrating with HTTP
[02:20:03] <marc.blanchet.qc> less code is not only saving various resources, but makes the implementation more robust and reliable, which is pretty key for embedded use cases.
[02:20:03] Kurt Andersen leaves the room
[02:20:31] <Edward Birrane> PIB not PID
[02:20:55] <Joseph> What is the goal of streaming with the Bundle Protocol?  They seem like two opposing concepts
[02:21:08] <Scott Burleigh> I am totally with you.  The dictionary, in particular, is a major threat to robustness and reliability.
[02:21:26] John Leslie leaves the room
[02:22:08] John Leslie joins the room
[02:22:22] <Henry Suthon> Streaming may be useful for storage constrained devices.
[02:23:13] <marc.blanchet.qc> streaming over dtn is being discussed in CCSDS.
[02:23:17] Dave Crocker joins the room
[02:24:11] <Scott Burleigh> I can amplify, if that's worthwhile.
[02:24:11] <Keith Scott> The streaming applications that NASA has worked with are targeted at getting 'best available' 'real-time' data as it becomes available, with the 'complete' (reconstituted, via custody transfer or other reliability mechanisms) stream reconstituted later.  Think of a DVR where if you're offline you see a black screen, but later you can 'rewind' and see the missing parts.
[02:25:25] <Keith Scott> With respect to the inclusion of time in Bundle Protocol: If all you have is hop count you have a dual problem with bundles that get to some location and can't move forward, and then stay there forever.  A combination of time and/or hop count seems most robust.
[02:25:43] <Scott Burleigh> Right, expiration time is our best defense against congestion.
[02:25:57] <Joseph> Support revisiting SDNVs too if that is in the charter
[02:26:11] <Keith Scott> In some environments, not all.  But in those environments, I think it's really important.
[02:26:59] <Joseph> If removing the dictionary is on the table, then I don't see how this topic of time can be considered different
[02:27:03] <Will Ivancic> Presentation is over
[02:28:09] <Joseph> I would like to see a study of what times were set in bundles and why
[02:29:24] <Scott Burleigh> I would say expiration time is architecturally fundamental while the dictionary is just a bandwidth optimization.  But it's worth talking about.
[02:29:32] <Edward Birrane> I don't have a study, but I certainly see different routing behavior in CGR based on expiration time.
[02:29:58] <John Dowdell> Fred Templin now presenting
[02:30:08] <Alex Amirante> Presentation is over
[02:30:15] <Joseph> That last comment was for all parties interested, not just the presenter
[02:31:45] Keith Scott leaves the room
[02:32:03] Keith Scott joins the room
[02:32:33] <Elwyn Davies> In terms of deciding what to drop when resources run out, the length of time that the bundle has been in the network is interesting as well as the expiry time.  Prsenttly you can find this out from DTN2 bundle IDs but it not in the protocol.
[02:34:52] <Elwyn Davies> There are clearly two factors in setting expiry times - the expected delay before the bundle might reach its destination (network factor) and the time when the information might no longer be of interest (application factor)
[02:36:06] <Edward Birrane> In sensor networks, uniqueness of data correlates to its importance. Sensors that constantly spam an average temperature, for example, can afford to lose several samples.
[02:36:13] <Keith Scott> I would like to propose that the WG consider as *a factor* the ability to gateway to and from RFC5050 (in an application-independent way).  That would ease transition for people / organizations that are starting to deploy RFC5050-based DTN.
[02:36:42] <Scott Burleigh> Elwyn, I agree.  The "network factor" maps to a lower limit, the "application factor" to an upper limit.
[02:38:03] <Keith Scott> There's an expired draft that we implemented at one time, the superseding bundle extension block, that let applications explicitly identify related data for the purpose of removing duplicates in the network.
[02:38:05] <Keith Scott> https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&cad=rja&uact=8&ved=0CB0QFjAA&url=https%3A%2F%2Ftools.ietf.org%2Fhtml%2Fdraft-parikh-bundle-superseding-extension-block-00&ei=WhlkVNCLFcrKPYvEgcgO&usg=AFQjCNFgrsdfYFHQNabMaDfKkwwU6E2nVg&sig2=d4xq6Kxrr8PXvbWVVzh7eA&bvm=bv.79189006,d.ZWU
[02:39:23] <Elwyn Davies> But how do you route it across all those pieces?
[02:40:41] <Keith Scott> yes: end-to-end integrity
[02:40:45] <Edward Birrane> Yes
[02:40:46] <Will Ivancic> I think header integrity
[02:40:56] <Will Ivancic> end to end payload is application problem
[02:40:58] <Scott Burleigh> We need a multi-domain routing concept.  I don't doubt for a moment that it's possible, though.
[02:41:28] <Edward Birrane> Question is whether end-to-end integrity is separate from security.
[02:41:29] <Elwyn Davies> How many people are in the room roughly, please?
[02:41:50] <jeferson.nobre> 50
[02:41:54] <Elwyn Davies> @edward: agreed
[02:42:15] <Keith Scott> yes: considering the implications of protocol decisions on constrained environments should be a consideration
[02:42:19] <Henry Suthon> Yes, constrained is of interest
[02:42:20] <Scott Burleigh> Can us remote people raise our hands somehow?
[02:42:21] <Will Ivancic> yes
[02:42:44] <Joseph> jeferson.nobre: Thanks for that data point, good to know how many folks are in the room
[02:43:00] <John Dowdell> say raise hand
[02:43:25] <Edward Birrane> I believe NM is required for operational deployment.
[02:43:38] <jeferson.nobre> now, I counted: 48
[02:43:42] <jeferson.nobre> :)
[02:43:44] <Will Ivancic> raise hand network management.
[02:43:44] <Elwyn Davies> NM: raise hand
[02:44:01] <Scott Burleigh> Raise hand.
[02:44:05] <Edward Birrane> Raise hand
[02:44:12] <Henry Suthon> raise hand for NM
[02:44:31] <John Dowdell> votes for no?
[02:44:35] <Elwyn Davies> NM: Does this include remote config?
[02:44:39] <jeferson.nobre> 1
[02:44:41] <Keith Scott> eed to specify addressing -- if we're going to work at all we have to do SOMETHING (either simply accept the current addressing scheme).
[02:44:56] <Scott Burleigh> Raise hand.
[02:45:04] <Keith Scott> or develop something new.
[02:45:06] <Will Ivancic> If we do not address ADDRESSING, we will not scale and therefore will not widely deploy.
[02:45:12] <Elwyn Davies> addressing: hand down
[02:45:42] <Keith Scott> raise hand for neighbor discovery / dynamic routing (node and contact discovery)
[02:45:45] <Scott Burleigh> Raise hand.
[02:45:49] <Anders Lindgren> +1
[02:45:56] <Elwyn Davies> discovery: raise hand
[02:46:05] <Will Ivancic> Getting into way to much work early on.  All need to be done.
[02:46:35] <Elwyn Davies> routing: Effectively out of scope for charter
[02:46:52] <Scott Burleigh> Not in this charter.
[02:47:01] <Elwyn Davies> routing: yes but out of scope by charter
[02:47:03] <Will Ivancic> Agreed, routing out of scope for charter as are many of these
[02:47:12] <Edward Birrane> ROuting: Out of scope. Recommend from RG
[02:47:24] <Anders Lindgren> routing: hand up as in I'm interested, but maybe in RG
[02:47:40] <Keith Scott> I think routing is a RG topic.
[02:47:57] <Elwyn Davies> There are already > 200 routing proposals!!!
[02:48:30] <Elwyn Davies> Essentially none of them are generally practical
[02:50:08] <Joseph> +1 for taking the list to the list
[02:51:44] tomek leaves the room
[02:51:48] <Keith Scott> thanks!
[02:51:51] <John Dowdell> jabber scribe has left the room
[02:51:59] Suzanne Davidson leaves the room
[02:52:01] david eddy leaves the room
[02:52:01] John Dowdell leaves the room
[02:52:04] <Scott Burleigh> Thanks, chairs and scribes.
[02:52:08] Will Ivancic leaves the room
[02:52:12] Anders Lindgren leaves the room
[02:52:19] Brian Haberman leaves the room
[02:52:21] Edward Birrane leaves the room
[02:52:25] jeferson.nobre leaves the room
[02:52:26] Scott Burleigh leaves the room
[02:52:28] Fuyou Miao leaves the room
[02:52:34] Meetecho feed leaves the room
[02:52:52] Joseph leaves the room
[02:52:54] Elwyn Davies leaves the room
[02:52:56] marc.blanchet.qc leaves the room
[02:52:59] Henry Suthon leaves the room
[02:53:05] John Dowdell joins the room
[02:53:05] John Leslie leaves the room
[02:53:12] Dave Crocker leaves the room
[02:53:12] John Dowdell leaves the room
[02:53:17] Keith Scott leaves the room
[02:53:22] Alex Amirante leaves the room
[02:53:27] Meetecho leaves the room
[02:53:37] Franck Martin leaves the room
[02:53:52] elwynd leaves the room
[02:54:48] Will Ivancic joins the room
[02:55:53] jlcJohn leaves the room
[03:02:00] Kurt Andersen joins the room
[03:02:02] Will Ivancic leaves the room
[03:17:08] Kurt Andersen leaves the room
[03:17:30] Kurt Andersen joins the room
[03:21:24] marc.blanchet.qc joins the room
[03:27:40] marc.blanchet.qc leaves the room
[03:28:35] tomek joins the room
[03:42:30] tomek leaves the room
[05:05:38] Kurt Andersen leaves the room
[15:37:26] tomek joins the room
[16:45:20] Kurt Andersen joins the room
[18:07:06] Kurt Andersen leaves the room
[19:09:47] tomek leaves the room
[19:30:09] tomek joins the room
[19:56:52] tomek leaves the room
[20:07:23] tomek joins the room
[21:02:10] tomek leaves the room
[21:16:10] tomek joins the room
[21:23:43] tomek leaves the room
[21:43:19] tomek joins the room
[21:44:32] tomek leaves the room
[23:11:49] tomek joins the room
[23:36:07] tomek leaves the room
Powered by ejabberd Powered by Erlang Valid XHTML 1.0 Transitional Valid CSS!