1. Sigma Xi Guest Lecture February 27, 2002 Survivable Networks: Protecting the Internet and phones from “backhoe-fades” and other hazards Wayne D. Grover University of Alberta, Dept. of Electrical & Computer Engineering TRLabs, Network Systems Group

2. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 2 Outline (selected topics) Background –Fiber Optics, DWDM, –Multiplexing & Concept of a Transport Network –Goals and Impacts of Protection / Restoration times Rings –types –multi-ring network design Span-restorable mesh –concept, self-organizing approach –capacity design Path-restorable networks p-cycles Some Current Research

3. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 3 US Circuit Switched Voice and Internet Traffic CAGR 1996-2005 Internet 95.8% Voice over IP 30% Data Traffic 30% Circuit Switched 12.1% Terabytes / day 0 2,000 4,000 6,000 8,000 10,000 12,000 14,000 1996199719981999200020012002200320042005 Source: Renaissance Analysis via Marconi PLC 2001

4. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 4 Fiber Optics and WDM: Wavelength (nm) 16001700 1400 1300 1200 1500 Attenuation (dB/km) 0.1 0.2 0.3 0.4 0.5 0.6 1310nm 1550nm

5. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 5 Dense WDM: ITU Channel Spacing 16001700 1400 1300 1200 1500 Attenuation (dB/km) Wavelength (nm) 0.1 0.2 0.3 0.4 0.5 0.6 1525 15301535 1540 15451550155515601565 ITU Channel Spacing And each wavelength can carry ~ OC-192 (10 Gb/s)

6. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 6 How important is one little fiber? If 64Kb/s = 1 lane Then Based on Current Technology, a single Fiber would = 25 Million Lanes, or a Highway that was 60,000 Miles Wide Adapted from Marconi OctoBrief 2001

7. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 7

8. 8 British Telecom

9. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 9

10. 10 The Level(3) N. American Network

11. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 11 32-node Italian backbone transport network some real network topologies

12. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 12 Belgian national transport network (Belga 39 - 39 nodes, 59 spans) some real network topologies

13. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 13 “COST 239” European Community project model ( 19 nodes, 40 spans) some real network topologies

14. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 14 “MCI” North American continental backbone (disguised topology only) some real network topologies

15. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 15 So everything is fine until.... “ Universal Cable Locator “ !!

16. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 16 News Reports Massive fiber cuts interrupt Net traffic –"Let me tell you, it really hurts right now," said AboveNet's chief technology officer. "We were given a 1 hour estimate for this problem to be corrected." SEA-ME-WE3 cable cut again –The cable was damaged by sand-mining operations in Indonesian territorial waters about 50 kilometers south of Singapore. International traffic from Australia was seriously affected. Massive Fiber Cut Pauses East-West Traffic –A fiber-optic cable cut in Ohio interrupted all forms of traffic across the United States for nearly 12 hours Wednesday has been repaired.... four OC-192 lines that were accidentally severed by a gas company employee digging with a backhoe.

17. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 17 Service Impact of failure duration 0.010.11101001K Service Impact Severity Time (sec) 10K 50 - 150 ms100s of ms10sec to mins15 min to 30 min Severe Business Impact: Regulatory reporting Application timeout, Business Impact TCP Session timeout, X25 disconnect TCP Backoff, unfairness, User terminates session, All voice calls lost Business Impact TCP re-transmit, minor delays, Some voice calls dropped, Video degradation No impact, TCP recovers, Reframe No impact, TCP recovers, 5% Voice disconnect 2 sec : all circuit-switched connections dropped Target range Impact (Log scale)

18. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 18 Concept of a “transport network” End-users Service layer Logical layer Physical layer system geographical

19. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 Network Layers Logical Trunk Group of n x DS1 OCn Switch DCS New York San Francisco Service Layer Transport Layer n x DS3 Switch

20. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 20 Concept of a transport network: one physical network - billions of logical network possibilities…through cross-connects

21. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 21 Optical cross-connect Optical transport system (1.55  m) Optical transport system (1.55  m) Fibers In Fibers Out -Mux Add ports Drop ports... Transparency = node-bypass Optical-layer Cross-connect (Optical or Electronic Fabric)

22. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 22 Chip size: 1 cm x 1 cm Source: L-Y. Lin (AT&T) Optical Layer Switching An 8x8 Switch

23. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 23 Each layer has a native form of “demand units” that are aggregated into capacity units of the next lower layer End-users Service layer Logical layer Physical layer system geographical Erlangs, packets, private lines, ATM VCs #s of: DS-0, DS-1, VPs, STS-n(PL), STSn(IP) #s of: OC-48, OC-192, wavelengths #s of fibers, wavelength regens, add-drop #s of cables, ducts, transponders, spectral allocations “the transport network”

24. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 24 Some basic approaches to network survivability … (APS systems) –1+1 –1:1 –1:N -> rings – UPSR: unidirectional path switched rings – BLSR: bi-directional line- switched rings -> mesh –span - restorable –path - restorable (shared backup path protection) -> p-cycles (ring-mesh hybrids) –based on access / core principles –based on forcer clipping principle

25. Introduction to ring types, sizing and loading

26. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 26 Unidirectional Path-switched Ring... Unidirectional - because in normal operation all working demand flows in one direction only. i.e.,A sends to B clockwise, B also sends to A clockwise Path-switched - because in restoration each receiver selects an alternate end-to-end path through ring, regardless of where actual break occurred. Two main types of “survivable ring”.... UPSR

27. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 27 Protection fibre Working fibre 1 2 3 4 5  Tail-end Switch UPSR Animation...

28. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 28 A D E B C A -> B B -> A UPSR (OPPR)...line capacity requirement Consider a bi-directional demand quantity between nodes A, B: d A,B. - A to B may go on the short route - then B to A must go around the longer route Thus, every (bi-directional) demand pair circumnavigates the entire ring. Hence in any cross section of the ring, we would find one unidirectional instance of every demand flow between nodes of the ring. Therefore, the line capacity of the UPSR must be: “ The UPSR must have a line rate (capacity) greater (or equal to) the sum of all the (bi-directional) demand quantities between nodes of the ring. “

29. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 29 Bi-directional Line-switched Ring...Principle of operation (“4-fibre” BLSR illustrated) Bi-directional - because in normal operation working demand flows travel in opposite directions over the same route through the ring Line-switched - because in restoration the composite optical line transmission signal is switched to the other direction around the ring (on the other fibre pair) specifically around the failed section. Two main types of “survivable ring”.... BLSR

30. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 30 Protection fibres Working fibres Loop-back 1 2 3 4 5  (4 fibre) BLSR Animation...

31. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 31 BLSR …(OPSR) line capacity requirement both directions of a bi-directional demand can follow the short (or long) route between nodes “Bandwidth reuse” The line capacity of the BLSR must be: Planning issues / inefficiencies: - better than UPSR for non-hubbed - capacity dependence on demand pattern - “stranded capacity” - span exhaust A D E B C A -> B B -> A “ The BLSR must have a line rate (capacity) greater (or equal to) the largest sum of demands routed over any one span of the ring. “

32. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 32 From preceding it is evident that BLSR demand-serving ability depends in general on the demand pattern. Some of the recognized tendencies in real demand patterns are: or “mesh” ideal case for BLSR perfect bw re-use BLSR much more efficient than UPSR no optimization required this is the general tendency in inter-city backbone network optimization of ring loading this is a fairly exact model for access ring applications BLSR efficiency = UPSR same basic “access” demand pattern but dual hubs employed for access survivability Effect of some demand patterns on BLSR

33. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 33 with perfect bw re-use BLSR gets proportionally better as ring size increases with perfect hubbing demand patterns, BLSR never has any advantage over UPSR in this range optimized BLSR loading (and ring selection) can give significant benefits over UPSR Total demand serving capability Effectiveness of BLSR relative to UPSR depends on demand pattern

34. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 34 1. Ring “Sizing” - CONTEXT: A number of demand pairs are to be served by a BLSR - QUESTION IS: What is the minimum line rate BLSR required? demands that must be served Required BLSR line capcity line rate = f (demands, routing in ring) Q. What is it that has to be optimally decided to minimize the required line rate ? i.e. (What do we have control over here?) A. for each demand: cw, or ccw ? BLSR related optimization problems

35. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 35 2. Ring “Loading” - CONTEXT: A number of demand pairs are to be served, but not necessarily all in same ring. i.e., there is a “pool” of outstanding demands to consider for selection into a given ring. - QUESTION IS: What is the maximum number of these demands that a BLSR with given capacity can serve? or... (alternate goal) Which set of demands (and routings) achieves greatest utilization of ring capacity? pool of demands needing to be served ? which demands to pick ? fixed ring capacity BLSR related optimization problems

36. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 36 Multi-Ring Network Design Problem Design Method Given Network topology Demand pattern Ring types Cost model Min-cost Design Ring Systems  Type  OC-n size  Topological layout  Glass-through locations Routing  Ring assignment  Inter-ring transit locations Subject to: All demands served Capacity constraints Max. ADMs per ring Inter-ring transit locations Partial add/drop constraints Matched-nodes requirements, etc.

37. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 37 Upper bound on number of ring candidates for each graph cycle: Every combination of 2, 3, 4....up to N nodes defines a prospective collection of active ADM nodes that could be grouped together to define one ring. Upper bound on the number of different multi-ring designs that exist: Every combination of 1, 2, 3, 4....up to some pre-determined maximum number of rings can be considered as a multi-ring design solution.. and... also multiply by the number of “ring technologies” being considered. On the complexity of multi-ring design

38. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 38 illustration: a 10 node network: 10 13 possible rings, 10 21 possible multi-ring networks (over 100 million years to evaluate all designs at 10 6 design evaluations / sec.) ! Question: How big is  ?

39. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 39 Concept (each follows in more detail): graph coverage: Balance Capture Span elimination Dual-ring interconnect transit sites glass-throughs....a set of rings that covers every edge of the graph. This is one class of ring network.....in a BLSR, how well are the w i quantities “balanced” ? (since the largest of them dictates the protection capacity).....to what extent does a given ring tend to serve demands that both originate and terminate in the same ring.....a multi-ring design may not “cover” all graph edges, if the working demands can take non-shortest path routes.....for the highest service availability, some demands may employ geographically redundant duplicate inter-ring transfers....not all nodes may be sites where demands can switch rings.....each ring needs ADMs where demands add / drop, but not elsewhere ( ~> Express rings etc.). Concepts and principles in multi-ring design

40. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 40 a set of rings that uses or overlies all edges of the physical facilities graph is called a “ring cover”. “Coverage-based” design is a special (simpler) case of multi-ring design. a three ring “cover” a single ring design that may also serve all demands example “span eliminations” “Graph coverage” and concept of span elimination

41. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 41 the primary gateway node has a 1+1 receive selection setup here. protected by BLSR line-loopback reaction in r1 protected by BLSR line-loopback reaction in r2 protected by 1+1 APS inter-ring setup Concept of dual-ring interconnect (DRI) “drop-and-continue” method for BLSRs (also called Matched Nodes arrangement)

42. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 42 RCG is a transformation of the graph that represents the opportunities to transition from ring to ring. example: with ring-set given, r1 is connected to r2 through only one node. For DRI routing, only the RCG edges with 2 or more parallel arcs are available for routing Ring Connectivity Graph (RCG) for routing through ring networks

43. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 43 SCIP = “span coverage IP” * Fixed charge and routing model Modular aggregating routing * Iterated greedy ring placement * Eulerian decomposition demand re-packing * Hierarchical Rings Tabu Search * * = techniques used in combination in RingBuilder™ Some Mathematical Tools and Approaches to multi-ring network design

44. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 44 RingBuilder.... Main User Interface

45. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 45 RingBuilder.... “Advisor” Mode

46. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 46 State of the art and Research Directions in Multi-Ring Network Design Solution Quality Model Accuracy Eulerian Ring Covers (Gardner et al., ‘94). Ring Coverage IP (Kennington, ‘97). RingBuilder™ (Slevinsky,Grover, ‘93) Net-Solver (Gardner et al., ‘95) Simulated Annealing (Roberts, ‘94). Hierarchical Rings (Shi,Fonseka, ‘96). Strategic Options (Wasem,Wu ‘91) Research Goals RingBuilder™ (Slevinsky,Grover, ‘95) CapacitatedMulti-technologyMulti-period Probabilistic Topology Tabu Search (Morley,Grover, ‘01)

47. Mesh-restorable Network Design

48. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 48 The concept and vision of distributed mesh restoration on-line simulation of “Selfhealing network” Tellium Corp. Key attributes: sharing of spare capacity over failure scenarios completely adaptive to current network state (network is the database) real time ( << 1 second) assurances of 100% restorability with theoretical minimum of spare capacity self-monitoring no central control (except for oversight) no global view databases of network state required no conventional inter-nodal signalling protocols; “self-organizing”

49. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 49 Basics of Mesh-restorable networks : SHN Protocol (28 nodes, 31 spans) span cut

50. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 50 SHN Protocol Overview Node states: –Pre-failure state –Sender state -----------> multi-index “forward flooding” –Chooser state ----------> initiates reverse linking / index –Tandem node state ---> forward flood competition – ---> reverse linking, cross-connection Key concept of a “statelet” –not inter-processor messaging –fixed fields, channel associated –space / location encodes problem information The SHN protocol is an event-driven finite state machine (FSM)

51. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 51 SHN Tandem Nodes Rules 1)Keep list of ports where precursor statelets are presently found and sort statelets by: –increasing repeat count –increasing number of the port where they appear 2)Replace precursors by better ones when better ones appear 3) Try as much as possible to re-broadcast statelets to all other spans 3a) When full re-broadcast is not possible, consider statelets in order of repeat count starting with the lowest values. 4) When complement statelet is received it is copied to the port of the precursor, all re-broadcast of forward flooding statelets for the corresponding index is stopped and a cross-connection is made After any of these events the rebroadcast pattern is revised to follow rule 3

52. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 52 Self-Organizing Networks: other applications The basic mechanism (search and formation of paths), also referred to as “Capacity scavenging” used for SHN can be adapted for other tasks: –Automated service paths provisioning (“broad-band dial-up”) –Network Audit (advance detection of restorability limitations and/or locations where capacity will soon be exhausted) –Improved restorability to complete node failures For more details, see: [1] W. Grover, “Self-organizing broad-band transport networks,” Proceedings of the IEEE, vol. 85, no. 10, October 1997.

53. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 53 Basics of Mesh-restorable networks : SHN Protocol (28 nodes, 31 spans) 30% restoration 70% restoration 100% restoration span cut

54. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 54 Basics of Mesh-restorable networks (28 nodes, 31 spans) span cut 40% restoration 70% restoration 100% restoration

55. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 55 Basics of Mesh-restorable networks Spans where spare capacity was shared over the two failure scenarios ?..... This sharing efficiency increases with the degree of network connectivity “nodal degree”

56. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 56 Consider two idealizations: –(1) restoration is “end node limited” i.e., the min cut governing restoration path number is at one or the other of the custodial nodes –(2) node has span degree d –(3) all working span capacities are equal at the node then:... OCX W W W W d spans in total if any one span fails, the total spare capacity on the surviving (d-1) spans must be >= to w. hence.... redundancy = (node) d A simple lower-bound on achievable redundancy

57. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 57 Basics of Mesh-restorable networks Mesh networks require less capacity as graph connectivity increases (sample result) ~ 3x factor in potential network capacity requirement

58. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 58 Where: S, c i, s i, w i are spans, costs and capacities P i is a set of “eligible routes” for restoration of span i is an assignment of restoration flow for span i to the p th eligible route encodes the eligible restoration routes: = 1 if span j is in the p th eligible route for restoration of span i Subject to: Restorability : Spare capacity : A basic model for spare capacity allocation

59. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 59 Understanding the span-restorable mesh spare capacity problem Network structure Failure scenarios Greatest requirement on all spans Total spare capacity (minimize) Failure scenarios Flows over eligible routes Flows simulta- neously imposed on any span All other spare capacities s i values Represented in the eligible route - defining information input

60. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 60 Threshold value ( for the network shown ) ( Total spare capacity, total number of eligible restoration routes ) Minimum spare Below the design threshold hop-limit, solution quality is affected. Above the threshold hop limit, computational difficulty grows unnecessarily How hop-limit affects complexity and solution quality in basic SCA

61. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 61 Other approaches and refinements studied Sakauchi - “cut oriented” formulation “Transportation-like” problem formulation Max-latching heuristic jointly optimized routing of working paths and spare capacity Modular transmission system capacities Economy-of-scale in cost-optimization Secondary optimization to maximize dual-failure restorability Secondary optimization to control optical path properties

62. path restoration

63. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 63 “path restoration:” what we mean The set of working paths severed by a span cut are restored by establishing a set of replacement paths end-to-end, simultaneously, between each O-D pair affected. The replacement paths are formed on-demand using only shared spare capacity (and possibly released working capacity (stub release).) –There is no dedicated reservation of a 1-for-1 backup path for each working path. Path restoration is equivalent to abandoning the damaged pre-failure paths entirely and rapidly re-provisioning new paths end-to-end. Path restoration distributes the impact of failures and the recovery effort more widely over the network as a whole and therefore generally permits greater efficiency in spare capacity design. The capacity design and real-time restoration problems for path restoration are considerably more complex than span-restoration –the fall-back to each O-D pair creates a capacitated multi-commodity max-flow problem. –issue of mutual capacity constraints very important

64. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 64 Comparative illustration of span versus path restoration Pre-failure 3 service paths

65. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 65 Failure occurs Comparative illustration of span versus path restoration All 3 service paths are lost until …

66. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 66 Span restoration reaction First look at a span restoration reaction … Note: example only, exact routes depend on working and spare capacities 3 service paths are lost failed working capacity restored by span restoration

67. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 67 A span restoration reaction …(2) Loopback / backhaul This restoration path could stop here

68. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 68 Now view a path restoration reaction... Same failure occurs

69. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 69 Path restoration reaction …with “stub release” Path restoration action Stub release

70. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 70 Stub release is an option / issue which does not exist in span restoration. From a capacity design standpoint it is preferable to have stub- release. From an operational viewpoint stub release complicates things: –a means of automatic signaling needed to rapidly release the surviving working “stub” capacities, AIS (Alarm inhibit signal) usually serves nicely for this, however –after physical repair, the reversion process is more complex. Notes about stub release in path restoration

71. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 71 * some variables become pre-computable parameters in the variations that follow Input data Intermediate (internal) variables Design output variables Cost of m th module size on span j. Set of all point to point demand quantities, indexed by r amount of demand on relation r Set of all spans between mesh cross-connection points Set of eligible working routes for relation r Encodes routes in = 1 if span j is in q th route for relation r Set of eligible restoration routes for relation r upon failure i. = 1 if span j is in p th route for relation r upon failure i Stub release quantity on span j from failure i Amount of demand lost on relation r for failure i No. of operating working and spare links (channels) on span j No. of modules of type m to install on span j for min cost Capacity of m th module size Working and restoration routing solutions N.B. “relation” = “OD pair” Parameters and variables in path-restorable capacity design (in the master formulation)*

72. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 72 Cost of modules of all sizes placed on all spans S. t. Defines the amount of damaged working flow for each relation under each failure scenario All demands must be routed Working capacity on spans must be adequate (1) (2) (3) Master formulation for path-restorable capacity design

73. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 73 With stub release Without stub release Restorability of working flows for each relation Spare capacity on spans must be adequate (see note on stub release) Modularity of installed capacity (4) (5) (6) (7) Master formulation for path-restorable capacity design (2)

74. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 74 O D Relation r Route q Failure span i Other span j Working flow g r,q Span j enjoys a stub release “credit” of spare capacity = g r,q for any failure on span i such that: Understanding how the formulation effects “stub-release”

75. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 75 If integer but non-modular capacity is desired: –change objective function to cost-weighted sum of spares (and / or working, if joint) –drop set M (the family of modularities), variables and constraint (7) If non-joint design is desired: –drop (1), (2), (3), and (6) –pre-compute all and as input parameters based on the pre-defined routing –pre-compute all stub-release quantities according to (6) If stub-release is not desired: –drop (6), i.e., set all = 0 Variations and options within the master formulation

76. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 76 joint optimized routing makes a large difference in span restoration Typical result comparing span and path- restorable network designs “non joint” “joint” designs joint-span is about as efficient as non-joint path joint design adds relatively little benefit to path restoration

77. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 77 Capacity Comparison of various schemes vs. Network Connectivity (Nodal Degree) x3

78. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 78 p-cycles: Background and Motivation “ Ring “ A. 50 msec restoration times B. Complex network planning and growth C. High installed capacity for demand-served D. Simple, low-cost ADMs E. Hard to accommodate multiple service classes “Mesh” F. Possibly up to 1.5 sec restoration times G. Simple, exact capacity planning solutions H. well under 100% redundancy I. Relatively expensive DCS / OXC J. Easy / efficient to design for multiple service classes “ Shopping list” : A, D, G, H (and J) please...keep the rest

79. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 79 Background - ideas of mesh “preconfiguration”

80. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 80 Restoration using p-cycles If span i fails, p-cycle j provides one unit of restoration capacity If span i fails, p-cycle j provides two units of restoration capacity i j i j

81. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 81 Optimal Spare capacity design with p-cycles

82. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 82 Optimal Spare capacity design - Typical Results i.e., “mesh-like” capacity

83. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 83 Understanding why (optimally planned) p-cycles are so efficient... 9 Spares cover 9 Workers 9 Spares cover 19 Workers Spare Working Coverage UPSR or BLSR p-Cycle …same spare capacity “the clam-shell diagram”

84. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 84 Another p-cycle example This 6-span p-cycle covers 4 x 2 + 6 = 14 working demands for each unit of spare capacity on itself Recent Theoretical results: (1)p-cycles are most efficient possible pre- configured structure. (2)up to S protection relationships per link in p-cycle, where S = # spans in cycle.

85. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 85 Where is it all going? Current Research: p-cycle networking concept: “Ring speed with mesh efficiency” Availability analysis of ring, mesh and p-cycle networks Ring-mesh hybrid networks Distributed pre-planning Fundamental topology design & evolution Ring to mesh evolution: “ring mining” strategy Capacity design theory for uncertain demand forecasts Optimal location of wavelength converters & regens Optimal design for multiple Quality of Protection classes Traffic and failure adaptive Self-organizing networks Controlled over-subscription of capacity (for IP / MPLS) meta-mesh design concept (for sparse graphs) “maintenance immunity”

86. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 86 Other info and resources web site: www.ee.ualberta.ca/~grover some popular / general articles: –“Self-organizing broad-band transport networks,” Proceedings of the IEEE, vol. 85, no. 10, October 1997. –"New Options and Insights for Survivable Transport Networks," IEEE Communications Magazine, January 2002. –W.D. Grover, “Network Survivability: A Crucial Issue for the Information Society,” IEEE Canadian Review Magazine, Summer 1997, pp. 16-21. EE 681 web site (password needed for lectures)

87. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 For related papers or further information or clarification on topics within this presentation please contact: Wayne D. Grover TRLabs / Univ. of Alberta 780 - 441 - 3815 grover@edm.trlabs.ca www.ee.ualberta.ca/~grover

88. Sigma Xi Guest Lecture “Survivable Networks” W.D. Grover, Feb. 27, 2002 88 People and Technology for the Future