MINUTE FROM ROLL WG - IETF 89 Updated with audio archives: http://www.ietf.org/audio/ietf89/ietf89-viscount-20140306-1520-pm2.mp3 http://ietf89.conf.meetecho.com/index.php/Recorded_Sessions AGENDA IETF89 ROLL: Slides: http://www.ietf.org/proceedings/89/slides/slides-89-roll-1.pdf 1) note well - http://www.ietf.org/about/note-well.html (2min, 00:00) 2) state of all drafts slide (5min, 02:00) a) draft-ietf-roll-applicability-home-building b) draft-ietf-roll-applicability-ami c) draft-ietf-roll-security-threats d) draft-ietf-roll-trickle-mcast e) draft-ietf-roll-terminology RFC 7102 3) state of all issues slide (3min, 08:00) 4) updates to milestones, schedule and practice. (5min, 11:00) 5) update on: draft-ietf-roll-applicability-home-building (10min,21:00) 6) update on: draft-ietf-roll-applicability-ami (10min, 31:00) 7) draft-ietf-roll-rpl-industrial-applicability detail the resolution (3min, 34:00) 8) updates on: draft-ietf-roll-applicability-template. (4min, 38:00) 9) mpl issue: draft-roux-roll-mpl-eval-00 (15min, 53:00) 10) mpl issue: draft-doi-roll-mpl-parameter-configuration (10min, 63:00) 11) new work: draft-ajunior-roll-energy-awareness (15min, 78:00) 12) Open floor. (12 minutes) Note well presented - please read Ines Robles is participating remotely from Mendoza (Argentina) Antonio Junior is going to present remotely. Slide 4: Agenda presented MPL, energy awareness (remote presenter), open issues Slide 5: State of Active Internet Drafts Applicability statements – most issues closed Have new open issues for security-threat which the document shepherd reported during LC, Security threats will be handled by Michael MCast submitted to IESG ROLL terminology is RFC 7102 Related document list presented Slide6: Related Internet-Drafts Ask for the intention of the document to the authors A number of them are going to be presented today, One of them may be will be an Independent submission A number of them, we think that they no fix with the charter, want to know the opinion of the WG, it could represent new work, we will be having a recharter or close discussion in Toronto Slide7-9: Open tickets 95% of the tickets are closed. Slide 10-11: Milestones MPL was never in charter - should it have been? Slide 12: Peter Van der Stok presents on ROLL applicability Home/Building Slide 13: Home and Building Control Mostly control traffic Lots of commonality between home and building Some differences - mainly commissioning Slide 14 : Sporadic Traffic What kind of traffic are you looking for? shows sporadic traffic pattern where nodes send sporadically to an edge router and go out for the maintaining of the control 5 or 6 hops are expected Slide 15 : Regular Traffic shows that there is also expected local traffic between nodes - 1 to 2 hops typically -regular traffic Slide 16: Multicast Traffic shows that there will be multicast traffic from a node, again typically 1 to 2 hops between Source and Destination Slide 17: MPL and RPL-P2P are dominant states MPL and RPL-P2P will therefore be dominant. Timeliness noted as an important factor as well as high repetition rates Slide 18: Two appendices Mentioned the two appendices about RPL shortcomings and Link behaviours Looking forwards for comments on ML. Slide 19: Daniel Popa presents on ROLL applicability AMI (Advanced Metering Infrastructure) Slides 20-22: give a summary of what has changed Alignment with applicability-template-04, added Smart Grid QoS requirements, updates on Smart Grid, AMI System, Layer2 features. Re-scope, Update RPL profile section Update 6LowPAN option section, Add description of security features. Slide 22 does not recommend the use of 6lowpan fragmentation Slide 23: Remaining work items - Security for initial and incremental deployment, speak with the person that did the secdir review to discuss which solution will satisfy his comments Section 10 other related protocols→ maybe an empty list, should not be empty. and MAC/PHY details Slide 24: Questions? Michael Richardson: I push this guys quite a lot to pick storing or not storing or publish a second document Q: from Oana Iova Jabber - why is fragementation not recommended? A: Use of IEEE 802.15.4g and other MAC/PHYs allow much larger packets up to 2048 octets, so it would not be needed Q: from Matt Gillmore - is the new template going to be used? A: from Michael Richardson - Yes, minimal changes so not a big issue Slide 25: Michael Richardson presents on ROLL applicability template The idea was to specify the things necessary in deployment’s basis Slide 26: Asked for a secdir review with respect to the security-thread draft, got the secdir review but the document is a table content, it was not understood in the required way, there is some information missing. Early February we had a couple of calls with people involved in the process, to figure out what exactly went wrong and what it is going on. Last week the meeting was with the reviewers, they are going to stick with the docs until the issues are solved. Notes that the document is a work-in-progress and never meant for publication. Also need to make it clearer what the relationship between the documents are. Slide 27: Diagram connecting the applicability-statements. Michael asks for help regarding statement which states that applicability document supplies answers to a list of questions, find the relation between the docs. Michael made a starting. RFC6550 is original document and is light on security considerations. There was a security framework document, which became a security threats document. The template and thus the applicability drafts refers to the threats document Note that ROLL applicability for industrial may be handled in 6tisch now Slide 28: Alexandru Petrescu (AP) presents on MPL performance evaluation - draft-roux-roll-mpl-eval The project tried to simulate 500 sensors to predict message propagation to other nodes in the network. Simulator is time sampling simulator which analyzes what every node in the network's state would be at a given instance. It is very difficult to practically build a 500 mode network in the lab. Slide 29 Simulator principles shown propagation model used by simulator The simulator models both proactive and reactive model. Q. from Ulrich Herberg - is there any loss taken into consideration between the nodes A. Yes, interference is considered, which constitutes loss Slide 30: Main Loop shows the simulator execution model Slide 31: Simulator graphical insight shows the GUI which shows the scattered nodes, where every node acts as MPL forwarder The aim was to send data as frequently as possible to as many nodes as possible. Modification of the MPL parameters was then done to see the different behavior Slide 32: Example of Evaluation Result shows a summary of results Simulator is java application Shows a live demo of the simulator running. Software is available on request. public it can not be download Q. from Matt Gillmore - what network MAC/PHY are you assuming? A. Nothing is specified - nothing is assumed Q. from Daniel Popa - struggling to interpret the results on slide 32 A. Defined an expansion factor in the draft - theoretical property Q. from Daniel Popa - why did you not use an existing simulator which is tried and tested? A. Because we like writing software - not aware of any simulator for trickle-mcast or MPL? (DP) - why not use NS? (AP) - does it have trickle? (MR) - it is extensible Q. from Kerry Lynn - what is the main thing you learned? A. What power is needed to send to so many nodes Slide 33: Yuskuke Doi present on ROLL MPL parameter configuration Draft was updated yesterday based on session with RFC editor Slide 34: I-D.roll-trickle-mcast-06- Section 5.4 From MPL draft, there is no mechanism for setting trickle parameters. Candidates for mechanism are e.g. preconfigure, DHCPv6 Slide 35: Option Format shows a DHCP option which can be used to configure the parameters Slide 36:Optional MPL Domain Address shows that with an optlen of 16 (MPL Forwarders) and 32 (MPL domain) information can be configured Slide 37: A Challenge: How to Describe Timers within Small Number of Bits? asks the question of how to describe timers in a small number of bits. Suggestion to use floating point. Signed values not necessary so wastage. Base 10 more convenient than base 2. Slide 38: Short Floating Point for Timers shows a proposal for a compact exponential representation for timers Slide 39: Questions? poses a number of questions - is DHCP a good proposal? Can we dynamically change parameters? (MR): good to know about deployments by email, I want get a feeling from the room, a hum, a hum if this work is interesting for the group to consider adopting, can I have a hum for yes, (some hums), can I have for no (no hums). People who will review this draft (fews hands up) Q. form DP - do you need dynamic configuration and if yes, need to consider DHCP A. Maybe we do. Application can move from one domain to another. (MR) tune without turning off then on again Q2. from DP: Can any commercially available DHCP server be used? A. Yes, if there is a way of adding binary data Q. from KL - Question re. whether we do MPL work at all in ROLL - so that should come first? A: Yes Q. from Mehdi Mani (MM) - Considered any other way? A. CoAP/MIB/SNMP could be use Q: from MM - Assume that a node does have a DHCPv6 client? A; Yes Comment from Tom Herbst - DHCPv6 is useful for distributing startup parameters so CoAP is too late Slide 40: Presentation (remote) from Antonio Junior on energy awareness applicability guidelines Slide 41: Introduction Slide 42: Alignment with the ROLL charter states intention of draft to contribute energy aware metrics to RPL protocol. New metric container type according to RFC6551 Slide 43: Energy-awareness metrics global applicability guidelines states use of metrics improve performance by 30% Slide 44: Main Design Aspects shows proposed metric container type -ENR- EFS metrics Slide 45: Using Energy Aware Metric with RPL Routing metric: Shortest path offering the shortest ranking of proposed metrics (ENR or EFS) Slide 46: Conclusions and Future Work RPL Implementation, a single document for RPL. Q. from MM - is your metric compatible with e.g. ETX/ETT? A. Yes Q. from Dominque Bartel (DB) - your metric is an estimate of how long it has been in low power mode. How different is it from E-E metric A. The metric is an estimate of the energy of the nodes and is more about how quickly the nodes consume energy and also helps to maximise the use of the nodes which are mostly in idle mode. (DB) - so is it siginifcantly different from remainining energy percentage? A. Yes. Q. from DP - Do we expect it can be used in a mix environment (battery and maina) and can we force battery powered nodes to be leaves? A. If most nodes are mains powered, difference is minimal so most beneficial to networks with mostly battery powered nodes Question from MR re. abandoning industrial applicablity. There have been no objections so assume it will be abandoned, the silence is taken as consent. Kerry Lynn - MPL is routing protocol agnostic; it does solve a problem for LLNs so it made sense to develop it here. HNCP in homenet could use MPL - author was not aware of it. MLD proxy was similar work but MPL is simpler so feeling is that it should carry on in ROLL. MR - So we would leave MPL in the working group and leave the charter untouched. Or someone comes along with something significantly more than e.g. Yusuke's document which may form another WG. Reminder from MR - if you think you have anything important to bring up, please do. Toronto will likely be the last meeting.