IP over WDM network

1. IP over WDM network Fang Yu 294 Class Presentation

2. Outline • History of WDM networks • Current Internet: Multi-layer protocol stack between IP and WDM layers • Future: IP directly over WDM • Challenge • Virtual Topology Reconfiguration • Multi-layer routing • One proposal: Optical Burst Switching technologies

3. History • In the late 70s • First fiber based optical transmission system • Before 1995 • Mostly a single high-speed optical channel • All multiplexing done in electrical domain(TDM) • 50Mb/s to 10Gb/s data services • After 1995 • WDM allows simultaneously transmitting multiple high-speed channels on different frequencies (Up to 160 wavelengths today) • 40G per l (OC768) • Total link capacity = 160 l *40G =6.4 Tbps

4. Current TypicalProtocol Stacks IP ATM SONET WDM WDM

5. 4E 4E 1/0 DCS 1/0 DCS CHCG LA 4E 4E 1/0 DCS 1/0 DCS LA ATM/IP 3/1 DCS ATM/IP 3/1 DCS CHCG ATM/IP 3/1 DCS ATM/IP 3/1 DCS DACS III DACS III CHCG LA DACS III DACS III PHNX LA CHCG Wavelength Path Crossconnect Hard- Wired LA LA OTS OTS OTS OTS OTS OTS PHNX CHCG (OTS: Optical Transport System) Proprietary (20-400 Gb/s) Wavelength Mux Section Crossconnect CHCG LA PHNX Transport Layer Model Service Layers “Packet” “Packet” “Packet” “Packet” DS1 (1.5 Mb/s) 3/1 DCS Layer DS3 (45 Mb/s) Core ATM/IPLayers 3/3 DCS Layer (DACS III) DS3 (45 Mb/s) SONET ADM Layer ADM ADM ADM ADM ADM ADM ADM OC48+ (2.5+ Gb/s) Media Layer Fiber Conduit/ Sheath

6. Disadvantage of Current Multi-layer Protocol Stack • Inefficient • In IP over ATM over SONET over WDM network, 22% bandwidth used for protocol overhead • Layers often do not work in concert • Every layer now runs at its own speed. So, low speed devices cannot fill the wavelength bandwidth. • When detecting of failure, different layers compete for protection • Optical layer detects failure almost immediately, restores error in 2us to 60ms • SONET layer detects failure in 2.3–100 us, restores error in 60 ms

7. Disadvantage of Current Multi-layer Protocol Stack (Cont) • Functional overlap: So many layers are doing the same thing • Routing • Protections • Slow speed • Electronic devices can not catch the transmission speed available at optical layer • Latencies of connection

8. Electronic Network Electronic Network O/E/O E/O E/O O/E/O O/E/O O/E/O O/E/O O/E/O E/O E/O Electronic Network Electronic Network Optical Core Historical Reason for Multi-layer • SONET over WDM • Conventional WDM deployment is using SONET as standard interface to higher layers • IP over ATM • IP packets need to be mapped into ATM cells before transporting over WDM using SONET frame • OEO conversions at every node is easier to build than all optical switch

9. Simplified Protocol Stacks? IP IP Frame Relay WDM-aware Electronic layer ATM SONET WDM WDM Current Typical Protocol Stack Simplified Protocol Stack

10. IP Directly Over WDM? • Establish high-speed optical layer connections (lightpaths) • IP routers connected through lightpaths rather than fiber

11. 3 2 3 2 WC No  converters With  converters 1 New request 1 3 1 New request 1 3 Challenge for IP over WDM network • WDM-aware Electronic layer • Reconfiguration and load balancing • Protection and restoration • Optical flow switching • Network management/control • Cross-layer optimization • Reconfigurable (within milli-seconds) OXC • Wavelength Converters

12. Virtual Topology Reconfiguration • Physical topology • Seen by optical layer • Virtual topology: a set of nodes interconnected by light-paths (wavelength) • Seen by electronic layer • Reconfigure of light-paths in WDM network by • Changing the light path connectivity between electronic switches • Tuning of the transmitter wavelength and the frequency-selective-switches A B A B C D C D

13. Fixed 0.1 Routing Reconfigurable Blocking Probability X6 Routing 0.01 WDM ring, 20 nodes one transceiver/node call BW = 1 wavelength 0.001 0.01 0.02 0.03 0.04 0.05 Call arrival rate Virtual Topology Reconfiguration(Cont.) • Enable network to dynamically response to changing of traffic pattern • Load balancing • Efficiency • Issues: • Time scale of changes • Triggered by what mechanisms • IP routing properties (e. g. stability)

14. Network control User 1 User 2 . . . . . . Router 1 Router 2 Router 3 WDM layer Multi-layer Routing • IP layer routing is the bottleneck of present Internet • Solution: Routing long duration flows at lower layers • Conventional packet routing • Optical bypass of intermediate routers for high volume traffic • End-to end (user-to-user) flow of entire file bypassing routers LIDS

15. Switching all the packets in optical layer? • Requires intelligence in the optical layer • Need to store packet during header processing • Optical buffers are extremely hard to implement • 1 pkt = 12 kbits @ 10 Gbps requires 1.2 s of delay => 360 m of fiber) • Optical Packet Switch still has a long way to go………………………

16. Various Optical Switching Technologies

17. Optical Burst Packet Switching • Retrospect the goal of IP over WDM: • Avoid electronic bottlenecks • Decrease the cost by simplifying the multiple layer architecture • OBS is one proposal of how to realize such a network

18. Optical Burst Switching • Resources are allocated using one way reservation • Sender sends a request • Sender sends burst without waiting for an acknowledgement of its reservation request • Switch does preparation for the burst when getting the request • Bursts can have variable lengths • Burst switching does not necessarily require buffering

19. Various OBSs • The schemes differ in the way bandwidth release is triggered. • In-band-terminator (IBT) – header carries the routing information, then the payload followed by silence (needs to be done optically). • Tell-and-go (TAG) – a control packet is sent out to reserve resources and then the burst is sent without waiting for acknowledgement. Refresh packets are sent to keep the path alive.

20. Main Characteristics of Optical Burst Switching • There is a time separation(offset time) between header and data • Header and data are usually carried on different channels • Header goes through sophisticated electronic processing • Data is kept in optical domain

21. Conclusion • Current IP over ATM over SONET over WDM network is inefficient and redundant • Future IP directly over WDM network • Advantages • Less latency • Automatic provisioning • Higher bandwidth utilization • Challenge of packet directly over WDM network • Optical buffer • Optical burst switch is one of the proposed techniques to IP over WDM network

22. Reference • John Strand, “Optical Networking and IP over Optical”, Feb 4, 2002 • Kumar N. Sivarajan, “IP over Intelligent Optical Networks”, Jan 5, 2001 • Gaurav Agarwal, “A Brief Introduction to Optical Networks”, 2001 • Yang Lihong, “Optical Burst Switching”, CMU networking seminar presentation • Vincent W. S. Chan, “Optical Networks: Technology and Architecture” • Eytan Modian, “WDM-Based Packet Networks”, IEEE Communication Magazine, March 1999 • Ornan (Ori) Gerstel, Rajiv Ramaswami,, “Optical Layer Survivability—An Implementation Perspective”, IEEE Journal on selected areas in communications, October 2000 • Eytan Modiano, Aradhana Narula-Tam, “Survivable lightpath routing:a new approach to the design of WDM-based networks”, IEEE JSAC,April 2002 • R. Ramaswami and K. N. Sivarajan, Optical Networks: A Practical Perspective, San Francisco: Morgan Kaufmann, 1998.