1 / 31

WP2 UPC Contribution to A2.2.1: Route Management

WP2 UPC Contribution to A2.2.1: Route Management. Route management. Network models Packet-switched/wavelength-switched model Routing models / Route management models Static routing model(s) [ETH, UPC, all partners] Combined intra- and inter domain routing model(s) [ETH, UPC]

Télécharger la présentation

WP2 UPC Contribution to A2.2.1: Route Management

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. WP2UPC Contribution to A2.2.1: Route Management NOBEL: Berlin May 18-19, 2004

  2. Route management • Network models • Packet-switched/wavelength-switched model • Routing models / Route management models • Static routing model(s) [ETH, UPC, all partners] • Combined intra- and inter domain routing model(s) [ETH, UPC] • Adaptive routing model(s) • Predictive routing model(s) • Multicast routing model NOBEL: Berlin May 18-19, 2004

  3. Optical Packet and Optical Burst vsWavelength Switched Model • Physical • Technological requirements – how advanced optical components are expected • Complexity of the hardware (node architecture) • Computational • Node control algorithms complexity • Routing algorithms complexity • Performance • Efficiency, network utilization • Flexibility • Data formats, bitrates, ... • Label switching paradigm – paths (connections) granularity, scalability NOBEL: Berlin May 18-19, 2004

  4. Optical Packet and Optical Burst vsWavelength Switched Model • QoS • Difficulty in quality guarantees • Hardware and control algorithms complexity • Network • Control Plane implementation • Signalization overhead • Adaptation to traffic demands • Interworking • With legacy networks – edge node operation complexity (adaptation, aggregation, ...) • Costs • Hardware (node), building of the network, ... NOBEL: Berlin May 18-19, 2004

  5. Physical – technological, hardware requirements Optical Packet and Optical Burst vsWavelength Switched Model NOBEL: Berlin May 18-19, 2004

  6. Computational complexity Optical Packet and Optical Burst vsWavelength Switched Model NOBEL: Berlin May 18-19, 2004

  7. Optical Packet and Optical Burst vsWavelength Switched Model Performance Network utilization, efficiency • Wavelength switching • Not dynamically adopted (in real time) to the actual traffic demands • Efficiency up to 9 times worse than in OBS/OPS, very high wavelength consumption • Medium blocking probability • Burst switching • Network utilization higher than at WS (due to statistical multiplexing in optical domain) • High blocking probability - optical buffers need for fine network performance • Packet switching • Very high network utilization (statistical multiplexing in optical domain) • Needs FDLs and WC’s for high performance (low PLR) • Even with FDLs, packet delay is low due to fast optical switching (without O/E conversion of packet payload) NOBEL: Berlin May 18-19, 2004

  8. Flexibility Optical Packet and Optical Burst vsWavelength Switched Model NOBEL: Berlin May 18-19, 2004

  9. QoS Optical Packet and Optical Burst vsWavelength Switched Model NOBEL: Berlin May 18-19, 2004

  10. Network aspects Optical Packet and Optical Burst vsWavelength Switched Model NOBEL: Berlin May 18-19, 2004

  11. Optical Packet and Optical Burst vsWavelength Switched Model Interworking With legacy networks,edge node operation complexity (adaptation, aggregation, ...) • Wavelength switching • Lack of aggregation problem • Adaptation only in physical layer (e/o or wavelength conversion) • Burst switching • Burst assembly problem • Packet switching • Necessity of adaptation the data coming from legacy network to optical packet payload field • Packets disordering problem NOBEL: Berlin May 18-19, 2004

  12. Optical Packet and Optical Burst vsWavelength Switched Model COSTS Hardware (node), building of the network • Wavelength switching • Lower costs then in OBS/OPS case • But very high wavelength consumption • Burst switching • May use cheeper low speed switching elements than in OPS • Costs of advanced optical components • Low wavelength consumption • Packet switching • Very high costs of advanced optical components (FDLs units (for buffering, synchronization), very fast tunable wavelenght converters, very fast switching elements, …) • Costs of high performance electronic control unit • Low wavelength consumption NOBEL: Berlin May 18-19, 2004

  13. Summary Optical Packet and Optical Burst vsWavelength Switched Model NOBEL: Berlin May 18-19, 2004

  14. Nobel-WP3 UPC worksOPS environment • Previous works - studies on contention resolution algorithms for a single switch • UPC contributions in asynchronous, variable length packets scenario • Next step - studies on routing strategies for a network scenario • Adaptive vs. Multipath • Per-packet vs. per-connection • Further works • QoS management taking into account previous results NOBEL: Berlin May 18-19, 2004

  15. Nobel-WP3 UPC worksOBS environment • Previous works • Studies on contention resolution algorithms for a single switch • Burst assembly mechanisms • Signaling protocols • Next step • Studies on the effectiveness of multi-domain contention resolution in a network scenario • Studies on different routing strategies for a networks scenario • Further works • QoS management taking into account the previous results NOBEL: Berlin May 18-19, 2004

  16. Route management • Network models • Packet-switched/wavelength-switched model • Routing models / Route management models • Static routing model(s) [ETH, UPC, all partners] • Combined intra- and inter domain routing model(s) [ETH, UPC] • Adaptive routing model(s) • Predictive routing model(s) • Multicast routing model NOBEL: Berlin May 18-19, 2004

  17. Combined Intra and Inter-Domain routing model • Our Research Focus is on QoS Routing (QoSR) in Optical Networks: • Dynamic Intra-AS QoS light-path provisioning (Optical QoS aware IGP) • Dynamic Inter-AS QoS light-path provisioning (Optical QoS aware EGP) • Coupling between both QoSR mechanisms NOBEL: Berlin May 18-19, 2004

  18. Combined Intra and Inter-Domain routing model • Research Goal: provide a combined Intra and Inter-AS QoS Routing model with the following characteristics: • Highly scalable • Resilience: survivability • Loop-free • Support for different CoS and Policy Control • Clear cut between QoS aware IGP and QoS aware EGP • Per-CoS fast re-route provisioning • Efficiency in terms of the trade-off between the updating frequency, and distributing and maintaining routing state information (inaccuracies) • Suitable signaling for QoS: requirements of the Control Planes for both routing protocols, IGP and EGP NOBEL: Berlin May 18-19, 2004

  19. Combined Intra and Inter-Domain routing model • Line of work: • Survey optical extensions to classical IGPs and EGPs • Development of Metrics and Routing Algorithms for both Intra and Inter-Domain QoS Routing • Efficient coupling between both Routing Algorithms • We also plan to carefully manage how traffic flows so that no starvation of best effort traffic occurs NOBEL: Berlin May 18-19, 2004

  20. Adaptive routing: analyzing the effects of flooding on global network performance • Routing and Wavelength assignment problem (RWA) • Not tractable problem, so divided into: • Routing sub-problem • Wavelength assignment sub-problem • Routing sub-problem • Static routing • Dynamic (adaptive) routing • Static routing: • Fixed-routing • Fixed-alternate routing • Does not consider network dynamics • Dynamic (adaptive) routing: • Adaptive shortest-path routing • Least Congested Path (LCP) • Includes network dynamics in the route selection NOBEL: Berlin May 18-19, 2004

  21. Adaptive routing: analyzing the effects of flooding on global network performance • Dynamic vs static? • Static routing is simpler and not so complex • Dynamic routing is more appropriate for high dynamic networks • Dynamic routing issues: • Route selection must be adapted to network dynamics • Flooding mechanism is required • Mainly for high dynamic networks • Is the network state databases information accurate enough? • Routing inaccuracy problem • Non-suitable path selection because of having inaccurate network state information NOBEL: Berlin May 18-19, 2004

  22. Adaptive routing: analyzing the effects of flooding on global network performance • Flooding mechanism • In an N nodes network, each change results in a N2 messages to be flooded • Leads to instability and scalability • There are not many contributions on optical networks • New techniques must be sought • Approaches could be based on: • Updating by time (hold-down timer) as an IP extension • Updating by number of network state changes • Updating by minimum number of available resources NOBEL: Berlin May 18-19, 2004

  23. Adaptive routing: analyzing the effects of flooding on global network performance • Routing inaccuracy problem: • Routing algorithms must reduce the impact of selecting routes based on inaccurate routing information • New routing algorithms must be sought • Not many contributions in optical networks: • Approaches based on: • Dynamic bypass concept (BBOR): • Rerouting through alternative pre-computed paths • Prediction (PBR): • Route decision according to a “novel” concept of predicted network state information • Simultaneously, flooding is “almost” removed • In short, efforts must be done to develop new adaptive routing mechanisms which include these factors in the route decision NOBEL: Berlin May 18-19, 2004

  24. Prediction Based Routing Usual Routing Algorithms need update messages with information about the network state Network state information is not accurate: - Aggregating information - Triggering of update messages - latency associated in flooded the update messages Routing Algorithms utilise inaccurate state information (RIP) NOBEL: Berlin May 18-19, 2004

  25. Prediction Based Routing • Idea: Source nodes can learn which is the better path and wavelength without update messages • Dynamic learning according to the routing information obtained in previous connections set-up. (Based in branch prediction) • For each wavelength on a path there is a prediction table, PT, to predict the possibility of blocking • For each wavelength on a path there is a history register, WR, with information about if in the last cycles the wavelength on that path has been used NOBEL: Berlin May 18-19, 2004

  26. Prediction Based Routing Index to access PT from wavelength register histories Prediction: Read two-bit counter value < 2 not blocked, value > 1 blocked NOBEL: Berlin May 18-19, 2004

  27. Count PT lambda 0 of SP1 Count PT lambda 1 of SP1 Count PT lambda N-1 of SP1 Count PT lambda 0 of SP2 Count PT lambda N_1 of SP2 ... ... >1 >1 >1 >1 <2 <2 <2 SP1 lambda 0 SP1 lambda 1 SP1 lambda N-1 SP2 lambda 0 SP2 lambda N-1 Prediction Based Routing • Update PT: PT are updated increasing counter if connection request is blocked and decreasing otherwise • Update WR: WR of the wavelength used is updated with 0, and the WR of the unused wavelength are updated with 1 • Prediction Algorithm: Two shortest path, SP1, SP2 and N wavelengths NOBEL: Berlin May 18-19, 2004

  28. Multicast approach in optical transport networks • Main idea: to optimize optical resources utilization • Lightpaths are established point-to-multipoint to overcome the mismatching between optical and client granularities • 1xN Splitters are placed at the optical terminations in order to extend the lightpath to N destinations (N=3 in the example) NOBEL: Berlin May 18-19, 2004

  29. 3 S 1 1 1 2 1 4 1 3 3 5 2 2 1 4 5 Multicast approach in optical transport networks • Example: When a connection from 1 to 3 is requested, the optical channel is transparently extended to nodes 4 and 5 (to allocating future connections from 1 to these nodes) Although resources are wasted firstly, they will be recovered in the future (when new connections from 1 to 4 or 5 arrive). NOBEL: Berlin May 18-19, 2004

  30. Multicast approach in optical transport networks • As it seems difficult to fill a lightpath with traffic generated by a single source to an only destination, the lightpath capacity will be better used if it collects traffic from this source to many destinations. • This will only be true if the granularity difference between lightpath and connections accomplish some constraints. • Some preliminary simulations show that the applied strategy can perform well under certain conditions. NOBEL: Berlin May 18-19, 2004

  31. Multicast approach in optical transport networks • Work Plan: • Start simulations to study the feasibility of the proposed strategy • Study how to physically implement the multicast approach • Find the ratio between granularities and optimal N • To analyze different algorithms to implement the multicast approach • Simulate different traffic patterns NOBEL: Berlin May 18-19, 2004

More Related