OPS: Optical Packet Switches

Size: px

Start display at page:

Download "OPS: Optical Packet Switches"

Drusilla Randall
5 years ago
Views:

1 OPS: Optical Packet Switches Hiroaki Harai National Institute of Information and Communications Technology Sep 8, 2006 Optical Network Testbeds Workshop 3

2 Why do we Need OPS? Internet Traffic in Japan: approx. 500 Gbps Peta-bps backbone future: doubled per year 500 Tbps in 10 years Electronic packet switch Year 2004: Throughput 640Gbps (16x40 Gbps) Lightpath networks Need fully meshed connections/ feasible? Need complex traffic engineering Important technology for bandwidth-assured applications OPS networks Provide extremely high-throughput Much larger bandwidth for switching (> 40 Gbps) O/E/O: 40Gbps 64 x 622 Mbps bus, SERDES May need MPLS-like control (labels can be merged) Important to ubiquitous society Sep 8, 2006 H. Harai (NICT) 2

3 Optical Packet Switching Data-path is all-optical (No O/E/O) Switch, Buffer Increase data bandwidth Label lookup (i.e. forwarding) Electronic parallel processing? Optical processing 1xN 1xNLabel Switch Switch 1xN 1xNLabel Switch Switch 1xN 1xNLabel Switch Switch 1xN 1xNLabel Switch Switch Nx1 Nx1 Buffer Buffer Nx1 Nx1 Buffer Buffer Nx1 Nx1 Buffer Buffer Nx1 Nx1 Buffer Buffer Optical payload payload Electronic serial payload payload Routing Make Make a routing routing table table for for forwarding procedure Forwarding Determine output output port port from from the the routing routing table table Scheduling Avoid Avoid packet packet collision Priority Priority control control Electrical Optical payload header payload header Switching Switch Switch the the packet packet to to the the appropriate port port Buffering Store Store the the packets packets in in appropriate time time Sep 8, 2006 H. Harai (NICT) 3

4 What should be Solved for OPS? OPS Increasing number of ports of optical switch Electronic: 16x16, 40Gbps 640Gbps Optical: 128x128, 160Gbps Tbps 25 Waves 500 Tbps Increasing speed of label lookup and buffer management Wire-speed operation Increasing number of labels looked-up Several thousands (New L2 possibility) More (L3 switching) Increasing buffer size At least tens of fiber-delay-lines Decreasing guard time between packets Several nanoseconds OPS Monitor/Analyzer Bit error / Optical packet error Under developing in NICT Sep 8, 2006 H. Harai (NICT) 4

5 Remaining Topics of This Talk OPS Prototype Optical label lookup Optical buffer Electronic buffer management Sep 8, 2006 H. Harai (NICT) 5

6 NICT s 40Gbps-based based OPS Experiment N. Wada, H. Harai, F. Kubota, OFC 2003 (no. FS7). Sep 8, 2006 H. Harai (NICT) 6

7 Optical Code based Ultra Fast Label Processing Packet format Payload data Header (label) Replace to the optical code (label) Optical label has different modulation format with payload data Optical label is physically distinguished from payload data Optical hardware based label processing is available Fully passive, ultra high-speed optical label processing Sep 8, 2006 H. Harai (NICT) 7

Wada, IEEE Photonic Tech. Lett., vol. 11, pp. 1689 1691, Dec.

8 Time Domain Optical Code Processing --Measured Waveform at 8-chip, 8 200Gchips/s Ref.) K. Kitayama, N. Wada, IEEE Photonic Tech. Lett., vol. 11, pp , Dec Auto-correlation 10ps/div. 10ps/div. Cross-correlation Sep 8, 2006 H. Harai (NICT) 8

9 Optical Fiber-Delay Delay-Line Buffer Different lengths of FDLs Need at least tens of FDLs H. Furukawa, H. Harai, N. Wada, N. Takezawa, K. Nashimoto, T. Miyazaki, A 31-FDL Buffer Based on Trees of 1x8 PLZT Optical Switches, to be presented at ECOC 2006, no. Tu4.6.5, Sep Control signal Optical packets Buffer Manager Nx(B+1) switch T 2T 3T 4T (B-1)T Discard Sep 8, 2006 H. Harai (NICT) 9

10 Optical FDL Buffering at 160Gbps Buffer 1 Scheduler 1 LN-SW LN-SW LN-SW LN-SW LN-SW LN-SW Output port Intensity (a.u.) Intensity (a.u.) Buffer 2 1 Switch 1 out Packet collision!! Switch 2 out Time (2µs/div) LN-SW LN-SW LN-SW LN-SW Buffer 1 out Avoidance of collision Buffer 2 out LN-SW LN-SW Output port Single Switch Double Switches 1 1 Time (2µs/div) Noise Source: N. Wada (NICT)

High-Performance Buffer Management for Optical Fiber-Delay Delay-Line Buffer Establish practical-scale high-performance management for FDL buffer (1) Develop buffer management by parallel and

11 High-Performance Buffer Management for Optical Fiber-Delay Delay-Line Buffer Establish practical-scale high-performance management for FDL buffer (1) Develop buffer management by parallel and pipeline processing For number of ports, time complexity of each processor is O(1) Parallel expansion of sequential (i.e. round-robin) scheduling N-times higher throughput than sequential scheduling (2) Confirm feasibility of support for 128x40Gbps packet switch by FPGA 8 times higher performance than ASIC based router (16x40Gbps) IP packet granularity (64byte or more; 10 Gpps), variable length (3) Prototyping 8-port buffer management system Parallel and pipeline buffer management (N =8) l1 l2 l3 l4 l5 l6 l7 l8 P12 P13 P14 P15 P16 P17 P18 P23 P24 P25 P26 P27 P28 P35 P36 P37 P38 P q P41 P42 P43 P44 P45 P46 P47 P cf) H. Harai and M. Murata, IEEE/ACM Transactions on Networking, Feb Sep 8, 2006 H. Harai (NICT) 8-port buffer management hardware 11

12 Performance Comparison IN/OUT ports NICT OPS Prototype 2 NICT s Top Data (As of Sep, 2006) 2 * Electronic Router 16 Bit rate 160 Gbps 160 Gbps 40 Gbps Label processing 800 Mpps/port 10 Gpps (at 40Gbps) 125 Mpps/port ** Scheduling 4 Mpps 10 Gpps 2 Gpps ** Buffer 2/port 31/port 16000/port * Can scale with nanosecond optical switches ** Estimated data: Assumption of wire rate processing of 40byte-packets Sep 8, 2006 H. Harai (NICT) 12

13 Conclusion We need high-throughput backbone network for ubiquitous society OPS will provide extremely high-throughput Switching bandwidth is not limited Buffer size is increasing Electronic scheduling is fast NICT has developed OPS but, Need more advanced devices (e.g., ns-switch) and systems Thank you for your attention Acknowledgment N. Wada, H. Furukawa of Photonic Network Group in NICT for valuable discussion, collaboration, and some slides in OPS Sep 8, 2006 H. Harai (NICT) 13

Why Data Center Requires Both, OPS and OCS!

Why Data Center Requires Both, OPS and OCS! K. Kitayama 1, Y.-C. Huang 1, Y. Yoshida 1, R. Takahashi 2, and A. Hiramatsu 3 1. Osaka University, Japan kitayama@comm.eng.osaka-u.ac.jp 2. NTT Device Technology