Named Data Networking Operational Aspects - IoT as a Use-case
Download as PPT, PDF1 like161 views
The document discusses Named Data Networking (NDN) and its operational aspects, including: 1. NDN claims several gains over IP such as lower latency, reduced congestion, and intrinsic security, but also has challenges like requiring router caches and upgrades. 2. NDN has been tested on experimental testbeds for both fixed and wireless environments, as well as for IoT applications using the NDN-RIOT implementation. 3. Performance evaluations show NDN can forward traffic at rates over 20Gbps but routing approaches need more study, especially for wireless networks. The FIT IoT Lab provides an experimental facility for further IoT research.
Named Data Networking Operational Aspects - IoT as a Use-case
- 1. 1 COPELABS, [email protected] Named Data Network Operational Aspects IoT as Use-case Rute Sofia ([email protected]) 2017.11.28 Sabbatical project in cooperation with Siemens AG CT, TUM (Chair of Network Architectures and Services)
- 2. NDN Operational Aspects Overview 2 COPELABS, [email protected] • NDN Claimed Gains and Pains over IP • Generic • Performance aspects: packet processing, routing, caching, naming • NDN Operational Status • The NDN testbed • Fixed and Wireless Environments • NDN for IoT • NDN RIoT • The FIT IoT Testbed • Alternatives • Summary and Next Steps • Claimed Gains and Pains, Summary • Testbed experimentation • Performance
- 3. NDN Claimed Gains and Pains 3 COPELABS, [email protected]
- 4. NDN Claimed Gains and Pains Claimed Gains, Issues, Open Questions 4 COPELABS, [email protected] • Claimed Gains over TCP/IP • Lower latency • Reduced congestion • Support for mobility • Intrinsic security • Can secure from the source • Issues • Routers need to have caches • Routers need to be upgraded to understand store-and-forward (be content-centric) • Questions Is there a real (KPI based) gain in relying on NDN? What is the operational status? (technology readiness and innovation) What about interoperability?
- 5. NDN Claimed Gains and Pains 1. Comparison with IP* 5*Saxena, Divya, Vaskar Raychoudhury, Neeraj Suri, Christian Becker, and Jiannong Cao. "Named data networking: a survey." Computer Science Review 19 (2016): 15-55.
- 6. NDN Claimed Gains and Pains 1. Comparison with IP* 6*Saxena, Divya, Vaskar Raychoudhury, Neeraj Suri, Christian Becker, and Jiannong Cao. "Named data networking: a survey." Computer Science Review 19 (2016): 15-55.
- 7. NDN Claimed Gains and Pains 1. Comparison with IP* 7 *Saxena, Divya, Vaskar Raychoudhury, Neeraj Suri, Christian Becker, and Jiannong Cao. "Named data networking: a survey." Computer Science Review 19 (2016): 15-55.
- 8. NDN Claimed Gains and Pains 1. Packet processing Aspects 8 COPELABS, [email protected] ACM SIGCOMM 2013: D. Oran et al., “NDN on a Router, Forwarding at 20Gbps and Beyond (NDN data plane based on hash tables) • Implementation of NDN data plane on an Intel Xeon-based like card, for Cisco ASR 9000 router (with a software-based Integrated Service Module) • Operational understanding of NDN limitations and advantages • Focused on caching and forwarding • Which elements impact packet processing? • FIB size: O(10^8) • PIT size: O(10^17) • Note: no realistic way to build a FIB/PIT yet, and PIT requires per-packet updates • Forwarding process • Interest Forwarding Process comprises (per-packet basis) • Content Store lookup • PIT lookup and delete • CS delete and insert (cache replacement) • IP forwarding just does reading on a per-packet basis • Packet format • Originally, binary and incurred 35% of overhead due to decoding • Currently, TLV (Type-Length-Value) /some discussion ongoing in the IETF, possibility to improve
- 9. NDN Claimed Gains and Pains 1. Packet processing Aspects 9 COPELABS, [email protected] ACM SIGCOMM 2013: D. Oran et al., NDN on a Router (NDN data plane based on hash tables) • Focused on Packet processing, input and output • Performance evaluated in a software-based implementation • Traffic used: internet HTTP traces as NDN load (13 million HTTP URLs from 16 IRCache traces*) • Routing table emulated as long-tail (model derived from URL Blacklist**) • Interest forwarding workload --> 3.87 FIB lookups • Single core and multi-core performance *ftp://ircache.net/Traces/DITL-2007-01-09/ - currently not available; direct contact with D. Oran ** http://urlblacklist.com/
- 10. NDN Claimed Gains and Pains 1. Packet processing Aspects 10 ACM SIGCOMM 2013: D. Oran et al., NDN on a Router (NDN data plane based on hash tables) Summarising: • FIB and RIB size is an issue • HT implementation shows promising results • Throughput minimally impacted by different sizes of FIBs and RIBs • Achieved performance of 8MPPS and the router can forward traffic at a rate of 20Gbps
- 11. NDN Claimed Gains and Pains 1. Packet processing Aspects 11 COPELABS, [email protected] Packet Format •https://tools.ietf.org/html/draft-irtf-icnrg-ccnxmessages-05 • Type-Length-Value Format derived from CCNx (XML format) / 64 k possible types • 16-bit Type • 16-bit Length • 0x1000 - 0x1FFF reserved for experimental use • No explicit guidance for encryption •https://named-data.net/doc/ndn-tlv/tlv.html • TLV helps for reduced packet size • If fragmentation is required, it can be used on a hop-by-hop basis
- 12. NDN Claimed Gains and Pains 1. Routing and Forwarding 12 COPELABS, [email protected] • NDN relies on separate routing/forwarding plane • Routing takes care of setting the topology and policies and handling their long-term changes, as well as for updating the forwarding table • Forwarding performs interface ranking and probing – different Strategies • Forwarding is “smart” (adaptable) • PIT state prevents loops and assists in measurement •NDN can rely on Distance-vector as well as Link-state approaches • Currently, following a link-state approach for fixed networks (OSPF based: NSLR) • FIB is used both on IP as on NDN • IP: searches for the longest-match to the destination address to get the next-hop in the FIB • NDN: searches the NAME to find the next-HOP(S) in the FIB and fetches data (not necessarily the closest copies) Relevant KPIs for Future Work (routing analysis) • CPU utilization • PIT size (total number of PIT entries) • Memory consumption (total memory consumed during named content routing) • Network utilization (total data transferred over the network) • Interest retransmission rate (total number of interest packets re-issued due to e.g., packet loss) • Time-to-completion: time it takes until a request is completed.
- 13. NDN Claimed Gains and Pains 1. NLSR** / IP approaches* 13 COPELABS, [email protected] Full explanation on routing approaches: https://named-data.net/wp-content/uploads/2017/09/routing_in_ndn-icn-china- 2017.pdf **NLSR: Named Data Link State Routing
- 14. NDN Claimed Gains and Pains 1. Routing Under Discussion 14 COPELABS, [email protected] • NLSR (link-state) • Hyperbolic Routing • Lowers the need for FIB (each node only requires neighbors coordinates) BUT • Delay is still large, and not sensitive to short time changes •Geo-hyperbolic Routing • Still exhibits large delays (with node increase and time increase) Currently, discussing regionalized geo-hyperbolic routing (similar to intra and inter domain splitting) Remarks for future work: • There is not a clear understanding concerning link-state vs. distance vector relevancy in NDN. Any IP scheme can be used, just requires adaptation. • For wireless, the current LS approaches are not adequate • OTHER Routing solutions: • COPELABS is currently working on the comparison of LS vs DV. • COPELABS is developing an extension of NSLR for opportunistic environments (based on eigencentrality of the nodes, derived from contextualization).
- 16. NDN Operational Status NDN Experimental Environments 16 COPELABS, [email protected] • NDN Testbed , Technology Readiness Level 6* • Fixed network – easy to set additional poles • Supported routing: NLSR, Hyperbolic routing • SOME applications •Wireless support (not integrated in the testbed) • NDN for Android, • NDN-Opp**, •IoT (not integrated) • RIOT-NDN • Emulator, mini-ndn*** • Full ndn network on 1 laptop *https://named-data.net/ndn-testbed/ **https://www.caida.org/workshops/ndn/1703/slides/ndn1703_sdynerowicz.pdf ***https://named-data.net/wp-content/uploads/2016/10/5-miniNDN.pdf
- 17. NDN Operational Status NDN for Wireless Environments 17 COPELABS, [email protected] • NDN for Android • Supports NDN in wireless (infrastructure) environments, via tunneling •NDN-Opp (NDN for opportunistic wireless environments)* • Supports NDN in wireless, infrastructure and D2D environments, both via tunneling and without IP • Supports also multihop wireless communication (no end to end path requirements) *http://copelabs.ulusofona.pt/index.php/research/projects/241-umobile
- 19. NDN for IoT Why? 19 COPELABS, [email protected] •Bring IoT Semantics to the network layers • Name Things and operations on Things • “Living room frontal view feed”, “CO level in kitchen” • “Living room frontal view feed”, “CO level in kitchen” • “max/min/avg pH of soil in specific point of US soil grid” • Focus on DATA associated with Things • Secure data directly • Latest updates, ACM ICN 2017 tutorial • http://conferences.sigcomm.org/ac m-icn/2017/files/tutorial-ndn- ccnlite-riot/1-ICN-intro.pdf
- 20. NDN for IoT Implementation: NDN RIOT* 20 COPELABS, [email protected] *http://conferences.sigcomm.org/acm-icn/2017/files/tutorial-ndn-ccnlite-riot/5-NDN-RIOT.pdf NDN RIOT is a project support by HAW Hamburg; INRIA, Florida International University; Zühlke GmbH
- 21. NDN for IoT Performance Aspects 21 COPELABS, [email protected] http://conferences.sigcomm.org/acm-icn/2017/files/tutorial-ndn-ccnlite-riot/5-NDN-RIOT.pdf
- 22. NDN for IoT Performance Aspects 22 COPELABS, [email protected]
- 23. NDN for IoT Performance Aspects 23 COPELABS, [email protected] http://conferences.sigcomm.org/acm-icn/2017/files/tutorial-ndn-ccnlite-riot/2-Why-ICN-for-IoT.pdf
- 24. NDN for IOT The FIT IOT Lab • Located in France several poles, 2732 wireless sensor nodes • Inria Grenoble (928), Inria Lille (640), ICube Strasbourg (400), Inria Saclay (307), Inria Rennes (256) and Institut Mines- Télécom Paris (160). • https://www.iot-lab.info/ • Part of OneLab 24 COPELABS, [email protected]
- 25. NDN for IOT The FIT IOT Lab – Tools Available • Web portal • Management and reservation tools (dashboard, nodes status, availability, statistics) • ssh • Access to open nodes and access to CLI • REST API (authentication service) • Access to services deployed by users 25 COPELABS, [email protected]
- 26. NDN for IOT Summary Claimed Gains and Pains, NDN vs. IP • From an architectural design perspective there seems to be clear benefits • Mobility and security, Support for intermittent connectivity • Focus on object names, and not on devices (more flexibility) • Performance: Only a Few Components have been analyzed • Packet processing in 1 router, implementation decisions on PIT and FIB • Routing: current solutions (link-state) still show delay; also not suitable for opportunistic wireless environments • Note: neither is IP, even though IPv6 shows better support Operational Status • Technology readiness level: 6 • International NDN testbed can assist in end-to-end experiments and in understanding core performance • Wireless environments require support by specific poles - the COPELABS NDN pole implements an opportunistic environment Interoperability • NDN achieves interoperability easily – currently, most implementations and NDN testbed rely on tunneling • Naming schemes require further analysis 26 COPELABS, [email protected]
- 27. NDN References COPELABS, [email protected] •Seyed Fayazbakhsh, Amin Tootoonchian, Yin Lin, Ali Ghodsi, KC Ng, Bruce Maggs, Vyas Sekar, Scott Shenker. Less Pain, Most of the Gain: Incrementally Deployable ICN. in SIGCOMM 2013. •Dirk Trossen and Alexandros Kostopoulos. Techno-Economic Aspects of Information-Centric Networking in Journal of Information Policy. •Bengt Ahlgren Information-centric networking and relaton to legal and regulatory issues by SAI •Yuan, T. Song, and P. Crowley. Scalable NDN forwarding: Concepts, issues and principles. In Proceedings of the 21st International Conference on Computer Communications and Networks , ICCN ’12, 2012. •u, W., & Pao, D. (2016). Hardware accelerator to speed up packet processing in NDN router. Computer Communications, 91, 109-119. •Saxena, Divya, Vaskar Raychoudhury, Neeraj Suri, Christian Becker, and Jiannong Cao. "Named data networking: a survey." Computer Science Review 19 (2016): 15-55. •IR Traces: ftp://ircache.net/Traces/DITL-2007-01-09/ - currently not available; direct contact with D. Oran •http://urlblacklist.com/ •Afanasyev, A. (2017). Named Data Networking of Things: NDN-RIOT, NDN-PI. Retrieved from http://conferences.sigcomm.org/acm-icn/2017/files/tutorial-ndn-ccnlite-riot/5-NDN-RIOT.pdf •Ahmed, S. H., & Kim, D. (2016). Named data networking-based smart home. ICT Express, 2(3), 130–134. https://doi.org/10.1016/j.icte.2016.08.007 •Aires, B. (2016). Design Principles for Named Data Networking. •Amadeo, M., Campolo, C., Iera, A., & Molinaro, A. (2014). Named data networking for IoT: An architectural perspective. EuCNC 2014 - European Conference on Networks and Communications, (October 2016). https://doi.org/10.1109/EuCNC.2014.6882665
- 28. NDN References COPELABS, [email protected] •Perino, D., & Varvello, M. (n.d.). A Reality Check for Content Centric Networking. Retrieved from https://pdfs.semanticscholar.org/cc28/2523f7123f8e7e7a4eae3aeb95d1aa3eca9b.pdf •Pesavento, D. (2016). Experimenting with NDN Apps using Mini-NDN. Retrieved from https://named- data.net/icn2016-tutorial •Saxena, D., & Roorkee, I. I. T. (2016). Named Data Networking: A Survey. Computer Science Review, Elsevier, 19, 15--55. https://doi.org/10.1016/j.cosrev.2016.01.001 •Shang, W., Afanasyev, A., & Zhang, L. (2016). The Design and Implementation of the NDN Protocol Stack for RIOT-OS. In 2016 IEEE Globecom Workshops (GC Wkshps) (pp. 1–6). IEEE. https://doi.org/10.1109/GLOCOMW.2016.7849061 •Shang, W., Bannis, A., Liang, T., Wang, Z., Yu, Y., Afanasyev, A., … Zhang, L. (n.d.). Named Data Networking of Things (Invited Paper). Retrieved from https://named-data.net/wp-content/uploads/2015/01/ndn-IOTDI- 2016.pdf •So, W., Narayanan, A., & Oran, D. (n.d.). Named Data Networking on a Router: Fast and DoS-resistant Forwarding with Hash Tables. Retrieved from https://pdfs.semanticscholar.org/dcd7/0db0da3a7d751dfde15e5df5086270799441.pdf •Zhang, L. (n.d.). Challenges in the Internet of Things Realization. Retrieved from https://named-data.net/wp- content/uploads/2016/07/challenges_iot_realization_pkujri.pdf
- 29. Name, e-mail