Understand p -Cycles, Enhanced Rings, and Oriented Cycle Covers

1. Understand p-Cycles, Enhanced Rings, and Oriented Cycle Covers Wayne D. Grover TRLabs and University of Alberta Edmonton, AB, Canada web site for related papers etc: http://www.ee.ualberta.ca/~grover/ ICOCN 2002, November 11-14, Singapore

2. Outline • What are p- Cycles ? • Why do we say they offer “mesh-efficiency with ring-speed ?” • Why are p-cycles so efficient ? • Comparison to rings and “enhanced rings” • Comparison to orientedcycle-covering techniques

3. The context • The domain for all that follows is the problem of network protection at the transport capacity layer. • i.e…. • Layer 3 inter-router lightwave channels • OBS-service layer working channels • Direct transport lighpaths • any other services or layers employing lightwave channels or paths All these sum to produce a certain number of working lightwavechannels on each span Philosophy:Protect the working capacity directly and it doesn’t matter what theservice type is

4. Rings... Fast, but not capacity - efficient

5. UPSR Animation... Working fibre 1 Tail-end Switch 5 2 Protection fibre 3 4 l1

6. UPSR (OPPR)...line capacity requirement A -> B A • Consider a bi-directional demand quantity between nodes A, B: dA,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 paircircumnavigates the entire ring. • Hence in any cross section of the ring,we would find one unidirectional instanceof every demand flow between nodes of the ring. • Therefore, the line capacity of the UPSRmust be: E B B -> A C D “ 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. “

7. (4 fiber) BLSR…(or OPSR) Working fibres 1 Loop-back 5 2 Protection fibres 3 4 l1 Loop-back

8. 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: A -> B A B -> A E B C D “ 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. “

9. A particular issue in multi-ring network design... Example of 3 (of 7) rings from an optimal design for network shown Ring 8 Ring span overlaps Ring 6 Ring 7 Ideally, BLSR-basednetworks would be 100% redundant. Span overlaps and load imbalances mean in practice they can be up to 300% redundant

10. Mesh... Capacity - efficient , but (traditionally argued to be) slower, and have been hampered by DCS / OCX port costs

11. span cut 30% restoration 70% restoration 100% restoration Concept of a span- (link-) restorable mesh network (28 nodes, 31 spans)

12. span cut 40% restoration 100% restoration 70% restoration Basics of Mesh-restorable networks (28 nodes, 31 spans)

13. Basics of Mesh-restorable networks Spans where spare capacity was shared over the two failurescenarios ? ..... This sharing efficiency increases with the degree of network connectivity “nodal degree”

14. ~ 3x factorin potentialnetworkcapacityrequirement Mesh networks require less capacity as graph connectivity increases

15. Now we also have “ p-cycles “.. “ p-cycles “.. Fast, and capacity efficient ....

16. Background - ideas of mesh “preconfiguration”

17. Protection using p-cycles If span i fails,p-cycle j provides one unit of restoration capacity i j If span i fails,p-cycle j provides two units of restoration capacity j i

18. Optimal Spare capacity design with p-cycles

19. i.e., “mesh-like” capacity Optimal Spare capacity design - Typical Results • “Excess Sparing” = Spare Capacity compared to Optimal Span-Restorable Mesh

20. Corroborating Results: COST239 European Study Network • Pan European optical core network • 11 nodes, 26 spans • Average nodal degree = 4.7 • Demand matrix • Distributed pattern • 1 to 11 lightpaths per node pair (average = 3.2) • 8 wavelengths per fiber • wavelength channels can either be used for demand routing or connected into p-cycles for protection Copenhagen London Berlin Amsterdam Brussels Luxembourg Prague Zurich Paris Vienna Milan

21. Corroborating Results... See: Schupke et al… ICC 2002 Schupke found p-cycle WDM designs could have as little as 34%redundancy for 100%span restorability

22. Understanding whyp-cycles are so efficient... Spare p-Cycle…with same spare capacity UPSR or BLSR Working Coverage 9 Spares cover 29 working on 19 spans 9 Spares cover 9 Workers “the clam-shell diagram”

23. Efficiency of p-Cycles (Logical) Redundancy = 2 * no. of straddling spans + 1* no. on-cycle spans ------------------------------------------------------------------ no. spans on cycle Example: 7 spans on-cycle, 2 straddlers : 7 / ( 7 + 2*2) = 0.636 Limiting case: p-cycle redundancy = N / ( N 2 - 2N)

24. The Unique Position p-Cycles Occupy Path rest, SBPP p -cycles: BLSR speedmesh efficiency Speed Span (link)rest. 200 ms BLSR “50 ms” UPSR 50 % 100 % 200 % Redundancy

25. ADM-like nodal device for p-cycle networking

26. Summary of Important Features of p-Cycles • Working paths go via shortest routes over the graph • p-Cycles are formed only in the spare capacity • Can be either OXC-based or on ADM-like nodal devices • a unit-capacityp-cycle protects: • one unit of working capacity for “on cycle” failures • two units of working capacity for “straddling” span failures • Straddling spans: • there may be up to N(N-1)/2 -N straddling span relationships • straddling spans each bear two working channels and zero spare • Only two nodes do any real-time switching for restoration • protection capacity is fully preconnected • switching actions are known prior to failure

27. Another recent development: --> “Enhanced Rings” ..and how they differ from p-cycles

28. To understand “enhanced rings..”consider If the fill level of the two “working fibers” at the span overlap is 50% each then the overall LA-SLC arrangement is 300% redundant ! i.e., (total protection + unused working) _________________________ used working

29. “Enhanced” rings... Idea is to allow the two “facing” rings to share switched access to a single common protection span. So, the cross-sectional view becomes:c Now, redundancy = 2 / 1 = 200%

30. Is an enhanced ring the same as a p-cycle ?... • No, because there is still a requirement for at least a matching amount of working and protection capacity on every span. • In other words protection is still only provided and used in the “on-cycle” ring-like type of protection reaction. • In contrast if the same problem is addressed with p-cycles, the troublesome span can be treated as: no protection fibers at all on straddling span: redundancy = 1 / 1 = 100% Or... no need to equip two working fibers if load does not require protection: redundancy ~ 0%

31. Another recent approach to reduce undesirable span overlaps in ring-based network design ... Oriented cycle double-covers

32. Bi-directional Cycle Covers • Consider the problem of “covering” all spans at a node with conventional bi-directional rings, without causing a span overlap... At an even degree node…there is no problem Even-degree node Odd degree node

33. Bi-directional Cycle Covers • Now consider the same problem of covering at an odd-degree nodec At an odd degree node…no bi-directional ring cover exists that does not involve a span overlap Even-degree node Odd degree node

34. But with Unidirectional (Oriented) Cycle Covers …you can always cover both even and odd nodes without the equivalent of a ring span overlap... examples of undirectional ring covers... Even-degree node Odd degree node (A mirror image set providesbidirectional W,P) The unidirectional ring coveravoids any double-coverage ! Equivalent to the bidirectional cover

35. So are Oriented Cycle Covers the same as p-cycles ? • No…because they still only protect in an on-cycle way. • The result is to get to ring-protection at exactly the 100% redundancy lower limit. • In an optimum oriented cycle cover every span will have exactly matching working and protection fibers. • P-cycles involve spans that have 2 working and zero protection fibers, which will never be found in an oriented cycle cover.

36. Summary • p-Cycles offer a promising new option for efficient realization of network protection • are preconfigured structures • use simple BLSR-like realtime switching • but are mesh-like in capacity efficiency • Other recent advances can be superficially confused with p-cycles: • enhanced rings reduce ring network redundancy by sharing protection capacity between adjacent rings • oriented cycle (double) covers adopt a undirectional graph cycle-covering approach to avoid span overlaps • Neither involves straddling spans; spans with working but no spare capacity • Both aim to approach their lower limits of 100% redundancy from well above 100% • p-cycles are well below 100% redundancy