INSTITUT FÜR INSTITUT FÜR NACHRICHTENVERMITTLUNG KOMMUNIKATIONSNETZE Universität Stuttgart Universität Stuttgart UND DATENVERARBEITUNG UND RECHNERSYSTEME Prof. Dr.-Ing. Dr. h. c. mult. P. J. Kühn Prof. Dr.-Ing. Dr. h. c. mult. P. J. Kühn Trends in Optical Burst Switching A Survey of OBS Research Christoph Gauger gauger@ikr.uni-stuttgart.de • Motivation and Introduction of OBS • Key Building Blocks and Selected Results • Trends and Viability E1 Workshop on OBS/OPS, Stuttgart, June 2004
OBS Design Rationale Internet emerged as the global platform for communication • Exploding traffic demand • Highly dynamic and asymmetric traffic profiles ➔ flexible and efficient transport network • QoS demanding applications ➔ transport network should offer QoS WDM transport introduced as cost-efficient transport layer • Increasing discrepancy between optical transmission and electronic switching speed ➔ keep data in optical layer • Flexible optical buffers difficult to realize ➔ avoid store-and-forward switching • No complex optical processing ➔ process control information electronically Institute of Communication Networks and Computer Engineering University of Stuttgart
OBS Design Rationale • OBS between packet and circuit switching - support IP traffic dynamics better than OCS - less complex optical layer than OPS switching required overprovisioning complexity for given IP traffic pattern granularity optical packet optical burst optical circuit switching switching switching Institute of Communication Networks and Computer Engineering University of Stuttgart
OBS Scenario . . . . . . . . . edge node core node OBS network control-channel OBS link data-channels • Burst assembly in edge node, • Fast optical switch mostly variable length • Separation of control and data • WDM-based transmission Institute of Communication Networks and Computer Engineering University of Stuttgart
OBS Building Blocks burst reservation burst scheduling . . . . contention . . resolution burst assembly QoS / service differentiation Definitions Burst assembly assembly of client layer data into bursts end-to-end burst transmission scheme Burst reservation assignment of resources in individual nodes Burst scheduling Contention resolution reaction in case of burst scheduling conflict Institute of Communication Networks and Computer Engineering University of Stuttgart
Burst Assembly control unit control data . . . . client layer data, optical bursts . . e.g., IP packets today 40…1500 Byte 10 kByte…10 MByte 32 ns…1.2 µ s at 10 Gbps 8 µ s…8 ms at 10 Gbps • Burst assembly triggered by time or size or both - burst arrival process - burst length distribution ➔ Strong impact on performance Institute of Communication Networks and Computer Engineering University of Stuttgart
Resource Allocation Burst reservation ∆ 1 Control Pretrans- ∆ 2 Offset Control mission ACK Data delay ∆ 3 Data one-pass reservation end-to-end setup • small vs. large pretransmission delay • blocking in core vs. at edge ➔ Determined by application scenario: network size and burst length Burst scheduling • Huge amount of proposals for optimized resource utilization • Void-filling Offsets produce voids ➔ void-filling algorithms - ➔ complexity of void-filling is not prohibitive (2 implementations reported) ➔ performance benefit sometimes overestimated Institute of Communication Networks and Computer Engineering University of Stuttgart
Contention Resolution • Burst loss possible due to bufferless statistical multiplexing • Application of OBS in high-speed metro/core networks ➔ lost data has to be retransmitted on end-to-end basis ➔ very low burst loss probability required (e.g., 10 -6 ) ➔ Need for highly effective contention resolution • Wavelength domain wavelength conversion - very effective as all WDM channels shared among all bursts ≥ λ s - but: low burst loss probabilities only for 100 ➔ additional schemes necessary • Time domain fiber delay lines (FDLs) • Space domain deflection/alternative routing • Segmentation only conflicting part of burst to be dropped ➔ Optimized combination of these schemes Institute of Communication Networks and Computer Engineering University of Stuttgart
Contention Resolution 0 0 10 10 0 10 8 FDLs -1 -1 10 10 no void filling, variable offset burst/packet loss probability burst/packet loss probability -2 -2 10 10 burst loss probability OPS, RNF -1 10 OPS, MINL -3 -3 10 10 OPS, MING -4 -4 10 10 void filling, variable offset -2 -5 -5 10 10 10 OBS, 4 FDLs OBS, 6 FDLs OBS, 8 FDLs -6 -6 10 10 -7 -7 10 10 -3 10 0.0 1.0 2.0 3.0 4.0 5.0 0 0 1 1 2 2 3 3 4 4 normalized basic buffer delay normalized basic buffer delay mean basic offset / mean burst transmission time • FDL buffer reservation in OBS and OPS - Different reservation strategies, early reservation with OBS - Joint work with Walter Cerroni in COST 266 • FDLs like offsets lead to reservations spread over time ➔ voids - void filling can reduce this negative effects - No improvement by void filling for offset == 0 or constant Institute of Communication Networks and Computer Engineering University of Stuttgart
Quality of Service QoS Differentiation Mechanisms Additional Preemption Intentional Resource QoS Scheduling QoS Offset (Segmentation) Dropping Reservation of Ctrl. Packets Requirements beyond differentiation capability • Robustness wrt/ network scenario and traffic characteristics • Low management complexity • Minimal processing and signalling effort Institute of Communication Networks and Computer Engineering University of Stuttgart
Node Design packet burst dynamic circuit Granularity edge delay Burst Assembly assumption: core rate approx. 10*access rate SOAs Switching MEMS Technology TWCs metro world End-to-end Signaling campus nation joint work 1 10 100 1 10 100 1 10 100 1 10 100 with HHI nano sec micro sec milli sec second • Granularity determines switching technology and vice versa - switching time << mean burst duration • Delay of 80km fiber • End-to-end delay constraint: few 100ms Institute of Communication Networks and Computer Engineering University of Stuttgart
Trends in OBS Future direction of OBS …more like OPS … more like OCS typical burst length short: 10…100 µ s, long: > 1 ms, some aggregation extensive aggregation burst reservation one-pass only one-pass or end-to-end contention resolution λ conv., FDL, deflection routing λ conv., defl./alternative routing … ➔ Viable if solutions are consistent: architecture + technology + control Institute of Communication Networks and Computer Engineering University of Stuttgart
Viability – Realization • OBS networks - benefit from aggregation and assembly - several architectural options available - can offer service differentiation to client layers • Technology - cost and availability of switching components still unsuitable - burst mode transmission requires changes in deployed infrastructure • Beneficial application scenario still open - core/transport vs. metro networks intensive traffic grooming towards core ➔ benefit of dynamic network? - • OBS has to fit into carriers’ world ➔ Interworking with circuit-switched photonic layer ➔ More effort towards efficiency, robustness, reduced complexity ➔ Application scenario and business model Institute of Communication Networks and Computer Engineering University of Stuttgart
INSTITUT FÜR INSTITUT FÜR NACHRICHTENVERMITTLUNG KOMMUNIKATIONSNETZE Universität Stuttgart Universität Stuttgart UND DATENVERARBEITUNG UND RECHNERSYSTEME Prof. Dr.-Ing. Dr. h. c. mult. P. J. Kühn Prof. Dr.-Ing. Dr. h. c. mult. P. J. Kühn Trends in Optical Burst Switching A Survey of OBS Research Christoph Gauger gauger@ikr.uni-stuttgart.de E1 Workshop on OBS/OPS, Stuttgart, June 2004
Provisioning Scenarios t setup week t setup t service = hour λ -channel second Flow millisec. t service Packet Burst nanosec. microsec. millisec. second hour week month year ≤ min. t setup t setup t service t setup t service « Institute of Communication Networks and Computer Engineering University of Stuttgart
Burst Assembly Can it reduce the detrimental impact of IP traffic characteristics? • Self-similarity on large time scales - early work suggested YES - recent publications prove NO for data plane • Smoothing on smaller time scales - consistent results show YES ➔ assembly really yields better performance Impact on TCP performance • In general positive due to smoothing • Assembly timer should be adapted to TCP congestion control Institute of Communication Networks and Computer Engineering University of Stuttgart
Burst Reservation source dest source dest ∆ 1 ∆ 1 Offset ∆ 2 ∆ 2 Pretrans- Control Request mission ∆ 3 ∆ 3 Data delay T p ACK Data t t one-pass reservation end-to-end setup • Burst loss in network • Burst loss at edge • Offset compensates processing • Dominated by propagation delay - Alternative: FDL in each node - long-haul networks (tens of ms) ➔ Mostly independent of network ➔ Only acceptable for large bursts size and burst length Institute of Communication Networks and Computer Engineering University of Stuttgart
Recommend
More recommend
