University of Alberta ECE Department Network Systems Gangxiang Shen, Wayne D. Grover Extending the p-Cycle Concept to Path-Segment Protection Gangxiang Shen, Wayne D. Grover URL:

University of Alberta ECE Department Network Systems Gangxiang Shen and Wayne D. Grover ICC2003, Anchorage, Alaska 2 Outline  Background and Motivation  Concept of Flow p-Cycles  Flow p-Cycle Design Model  Test Methods and Results  Operational Aspects and Potential Applications  Conclusions

University of Alberta ECE Department Network Systems Gangxiang Shen and Wayne D. Grover ICC2003, Anchorage, Alaska 3 Basic Approaches to Transport Network Survivability  p-Cycle 3Span p-cycles (since 1998) 3Flow p-cycles (our topic)  Mesh 3Span 3Path 3SBPP Efficiency Speed  Ring UPSR 3BLSR

University of Alberta ECE Department Network Systems Gangxiang Shen and Wayne D. Grover ICC2003, Anchorage, Alaska 4 Background: Span-Protecting p-Cycles  Characteristic: Ring-like switching speed and mesh-like capacity efficiency Straddling span On-cycle span

University of Alberta ECE Department Network Systems Gangxiang Shen and Wayne D. Grover ICC2003, Anchorage, Alaska 5 Comparison between Ring and p-Cycle Protection Ring network : p-Cycle: Spare Capacity Protection Coverage Able to restore 9 spansAble to restore 19 spans

University of Alberta ECE Department Network Systems Gangxiang Shen and Wayne D. Grover ICC2003, Anchorage, Alaska 6 The Unique Position Span p- Cycles Occupy Redundancy Speed “50 ms” 100 %50 %200 % Path rest, SBPP Span (link) rest. BLSR 200 ms p -cycles: BLSR speed mesh efficiency

University of Alberta ECE Department Network Systems Gangxiang Shen and Wayne D. Grover ICC2003, Anchorage, Alaska 7 Motivation  All the studies so far on p-cycles consider “span-protecting” p-cycles, so it is natural to ask: Q. is there is “a path protection equivalent to p-cycles?” -- A. Yes the answer is “Flow p-Cycles” !

University of Alberta ECE Department Network Systems Gangxiang Shen and Wayne D. Grover ICC2003, Anchorage, Alaska 8 Concept of Flow p-Cycles  Characteristic: Protect spans that span p-cycles can protect as well as spans that span p- cycles cannot protect (example: span 6-7 below) Intermediate node failure restoration (example: node 7) Path restoration –like spare capacity efficiency, 1:1 path protection –like switching speed On-cycle span Straddling span ? ?

University of Alberta ECE Department Network Systems Gangxiang Shen and Wayne D. Grover ICC2003, Anchorage, Alaska 9 The Position Flow p-Cycles Occupy Redundancy Speed “50 ms” 100 %50 %200 % Path rest, SBPP Span (link) rest. BLSR 200 ms p -cycles: BLSR speed mesh efficiency Flow p-cycles

University of Alberta ECE Department Network Systems Gangxiang Shen and Wayne D. Grover ICC2003, Anchorage, Alaska 10 Various Flow-to-Cycle Relationships Related basic concepts Intersecting and non-intersecting Intersection nodes Intersection flow segment Straddling and on-cycle flow relationship

University of Alberta ECE Department Network Systems Gangxiang Shen and Wayne D. Grover ICC2003, Anchorage, Alaska 11 Mutual Capacity Consideration  Single span-failure causes multiple flow- failures simultaneously  Flow-based restoration is required

University of Alberta ECE Department Network Systems Gangxiang Shen and Wayne D. Grover ICC2003, Anchorage, Alaska 12 Flow p-Cycle Design Model for 100% Span Failure Restoration  Objective: minimize total spare capacity  Constraints:  Affected flows upon a span failure must be fully restored  Number of cycle copies to build is set by the largest span failure-specific simultaneous use for unit copies of cycle  The spare capacity on a span must be enough to support the number of copies of each p-cycle that overlies the span

University of Alberta ECE Department Network Systems Gangxiang Shen and Wayne D. Grover ICC2003, Anchorage, Alaska 13 Test Networks

University of Alberta ECE Department Network Systems Gangxiang Shen and Wayne D. Grover ICC2003, Anchorage, Alaska 14 Result: Performance Comparison between Various Protection Schemes

University of Alberta ECE Department Network Systems Gangxiang Shen and Wayne D. Grover ICC2003, Anchorage, Alaska 15 Operational Aspects and Protocol

University of Alberta ECE Department Network Systems Gangxiang Shen and Wayne D. Grover ICC2003, Anchorage, Alaska 16 Applications of the General Concept

University of Alberta ECE Department Network Systems Gangxiang Shen and Wayne D. Grover ICC2003, Anchorage, Alaska 17 Conclusions  Flow p-cycle concept was proposed and evaluated  Flow p-cycle method achieves path restoration –like spare capacity efficiency and 1:1 path protection –like restoration speed

University of Alberta ECE Department Network Systems Gangxiang Shen and Wayne D. Grover ICC2003, Anchorage, Alaska 18 Future Work  Identify the impacts from the network details and demand patterns  Further consider operational aspects and develop control protocol  Implement some applications of the general concept  Consider an evolutional scheme, pre-configured segments: p-segments  Compare to “ordinary” node-encircling p-cycles for node protection.

University of Alberta ECE Department Network Systems Gangxiang Shen and Wayne D. Grover ICC2003, Anchorage, Alaska 19 Main References  [1] W. D. Grover and D. Stamatelakis, “Cycle-oriented distributed preconfiguration: Ring-like speed with mesh-like capacity for self-planning network restoration,” in Proc. of IEEE ICC’98, 1998, pp  [2] D. Stamatelakis, W. D. Grover, “IP layer restoration and network planning based on virtual protection cycles,” IEEE Journal on Selected Areas in Communications, vol.18, no.10, October 2000, pp  [3] D. A. Schupke, C. G. Gruber, and A. Autenrieth, “Optimal configuration of p- cycles in WDM networks,” in Proc. of IEEE ICC’02,  [4] W. D. Grover, and J. E. Doucette, “Advances in optical network design with p- cycles: Joint optimization and pre-selection of candidate p-cycles,” to appear in Proc. of the IEEE-LEOS Summer Topical Meeting on All Optical Networking,  [5] M. Herzberg and S. Bye, “An optimal spare-capacity assignment model for survivable network with hop limits,” in Proceedings of IEEE GLOBECOM’94, 1994, pp  [6] R. R. Iraschko, M.H. MacGregor, and W.D. Grover, “ Optimal capacity placement for path restoration in STM or ATM mesh-survivable networks,” IEEE/ACM Transactions on Networking, vol. 6, no. 3, June 1998, pp

University of Alberta ECE Department Network Systems Gangxiang Shen and Wayne D. Grover ICC2003, Anchorage, Alaska 20 Literature Survey on Span- Protecting p-Cycles  1998, Grover and Stamatelakis first proposed p-cycles concept and developed self-organized protocol [1]  2000, application of p-cycles to IP/MPLS layer [2] including node-encircling p-cycles  2002, application to DWDM networks [3]  2002, studies on joint optimization of p-cycle network designs [4]  ……

University of Alberta ECE Department Network Systems Gangxiang Shen and Wayne D. Grover ICC2003, Anchorage, Alaska 21 Evolution of Survivability Schemes  First Generation Pre-configured dedicated protection facilities  Fast restoration speed  Bad spare capacity redundancy Example: various ring-based techniques like 1+1, UPSR, BLSR  Second Generation Pre-planned but not pre-configured protection routes and shared spare capacities  Good spare capacity redundancy  Slow restoration speed Example: mesh-based restoration schemes like span, path, SBPP

University of Alberta ECE Department Network Systems Gangxiang Shen and Wayne D. Grover ICC2003, Anchorage, Alaska 22  Third Generation Pre-configured cycles  Fast restoration speed  Good spare capacity redundancy Example: p-cycles  Future Generation Pre-configured segments  Fast restoration speed  Good spare capacity redundancy Example: p-segments Evolution of Survivability Schemes (con’t)

University of Alberta ECE Department Network Systems Gangxiang Shen and Wayne D. Grover ICC2003, Anchorage, Alaska 23 Comparison between Mesh (SBPP) Restoration and p-Cycle Protection Mesh restorable network : p-Cycle: Two-way “talk” Generalized adaptive reconfiguration No two-way “talk” Immediate action Fully pre-prepared action On-cycle span PATH message RESV message

University of Alberta ECE Department Network Systems Gangxiang Shen and Wayne D. Grover ICC2003, Anchorage, Alaska 24 Cycle Preselection Strategy D is set of nonzero demand pairs (i.e., flows) on the traffic matrix S r is set of spans traversed by the working path between demand pair r P(j) denotes cycle j in cycle set P k enumerates spans on cycle P(j) and c k represents the cost of span k g r denotes number of traffic demand units of the working flow between demand pairs r l r denotes length of the working flow of demand pair r

University of Alberta ECE Department Network Systems Gangxiang Shen and Wayne D. Grover ICC2003, Anchorage, Alaska 25 Flow p-Cycle Design Model II: 100% Span and Node Failure Restoration  Objective: minimize total spare capacity  Constraints:  Affected flows upon a span failure must be fully restored  Affected flows upon an intermediate node failure must be fully restored  Number of cycle copies to build is set by the largest span failure- specific simultaneous use for unit copies of cycle  Number of cycle copies to build is set by the largest node failure- specific simultaneous use for unit copies of cycle  The spare capacity on a span must be enough to support the number of copies of each p-cycle that overlies the span

University of Alberta ECE Department Network Systems Gangxiang Shen and Wayne D. Grover ICC2003, Anchorage, Alaska 26 Concept of Multi-QoP  R 0 restorability: this is a wholly best-effort class with no assured restorability  R s restorability: this class is assured by design of restorability against any span failure, but receives only best efforts with no guarantee for node failure  R s+n restorability: this class enjoys assured restorability against any span failure or failure of an intermediate node other than its own end-nodes.

University of Alberta ECE Department Network Systems Gangxiang Shen and Wayne D. Grover ICC2003, Anchorage, Alaska 27 Flow p-Cycle Design Model III: Design to Support Multi-QoP  Objective: minimize total spare capacity  Constraints:  Affected R s and R s+n flows upon a span failure must be fully restored  Affected R s+n flows upon an intermediate node failure must be fully restored  Number of cycle copies to build is set by the largest span failure- specific simultaneous use for unit copies of cycle  Number of cycle copies to build is set by the largest node failure- specific simultaneous use for unit copies of cycle  The spare capacity on a span must be enough to support the number of copies of each p-cycle that overlies the span

University of Alberta ECE Department Network Systems Gangxiang Shen and Wayne D. Grover ICC2003, Anchorage, Alaska 28 Flow p-Cycle Design Model IV: Maximal Node Recovery under a Spare Capacity Budget  Objective: maximize overall node failure restorability  Constraints:  Affected flows upon a span failure must be fully restored  The restored traffic flows of a demand pair never exceeds its total lost traffic flows upon an intermediate node failure  Number of cycle copies to build is set by the largest span failure- specific simultaneous use for unit copies of cycle  Number of cycle copies to build is set by the largest node failure- specific simultaneous use for unit copies of cycle  The spare capacity on a span must be enough to support the number of copies of each p-cycle that overlies the span  Total spare capacity should not exceed a budget

University of Alberta ECE Department Network Systems Gangxiang Shen and Wayne D. Grover ICC2003, Anchorage, Alaska 29 Result II: Impact of Cycle Preselection Strategies and Number of Cycles of Flow p-Cycles

University of Alberta ECE Department Network Systems Gangxiang Shen and Wayne D. Grover ICC2003, Anchorage, Alaska 30 Result III: Impact of Physical Distance Limit of Cycle Circumference of Flow p- Cycles

University of Alberta ECE Department Network Systems Gangxiang Shen and Wayne D. Grover ICC2003, Anchorage, Alaska 31 Result IV: Performance on Node Failure Recovery (1) NetworksNSFNETARPA2SmallNet Intrinsic node failure restorability 98.32% 97.26%80.51% Redundancy increase (R s+n vs. R s ) 0.5%4.8%10.4% Total cost increase (R s+n vs. R s ) 0.2% 2.5% 3.6% R s : 100% span failure restoration R s+n : 100% span and intermediate node failure restoration

University of Alberta ECE Department Network Systems Gangxiang Shen and Wayne D. Grover ICC2003, Anchorage, Alaska 32 Result IV: Performance on Node Failure Recovery (2) Normalized spare capacity for 100% span failure restoration

University of Alberta ECE Department Network Systems Gangxiang Shen and Wayne D. Grover ICC2003, Anchorage, Alaska 33 Result IV: Performance on Node Failure Recovery (3) 100% R s services 100% R s+n services

University of Alberta ECE Department Network Systems Gangxiang Shen and Wayne D. Grover ICC2003, Anchorage, Alaska 34 Publications  [1] Wayne D. Grover, Gangxiang Shen, "Extending the p-cycle concept to path-segment protection," to appear in ICC2003, Anchorage, Alaska, USA, May,  [2] Gangxiang Shen, Wayne D. Grover, "Capacity requirements for network recovery from node failure with dynamic path restoration," to appear in OFC2003, Atlanta, Georgia, USA, March,  [3] Gangxiang Shen, Wayne D. Grover, “Extending the p-cycle concept to path-segment protection for span and node failure recovery,” submitted to IEEE JSAC special issue (Optical communications and networking series 2003).