Capacity Comparison of Mesh Network Restoration and Protection Schemes Under Varying Graph Connectivity John Doucette Wayne D. Grover TRLabs and University

Capacity Comparison of Mesh Capacity Comparison of Mesh Network Restoration and Protection Network Restoration and Protection

Schemes Under Varying Graph Schemes Under Varying Graph ConnectivityConnectivity

John DoucetteJohn Doucette

Wayne D. GroverWayne D. GroverTRLabs and University of Alberta

Edmonton, Canada

Copyright © W. Grover (TRLabs) November 2001

John Doucette, Wayne D. GroverCopyright © W. Grover (TRLabs) November 20012

Capacity Comparison of Mesh Network Restoration and Protection Schemes Under Varying Graph Connectivity

OutlineOutline

• Background• Restoration / Protection Schemes• Test Network Families with Varying Nodal Degree• Computational Aspects• Results• Interpretations and Summary• Further Reading



BackgroundBackground

• Growing number of restoration and protection mechanisms and design schemes– we identify six alternatives

to compare

• How does graph connectivity affect mesh-survivable capacity design?– Not widely studied aspect of

mesh networks– greater understanding can

guide topology planning– research method can vary

nodal degree as a parameter

d = 2.3

TORINO

GENOVA

ALESSANDRIA

PISA

MILANOBRESCIA

SAVONA

BOLOGNA

VERONA

VICENZA

VENEZIA

FIRENZEANCONA

PESCARA

PIACENZA

MILANO2

PERUGIA

L’AQUILA

ROMA

ROMA2

NAPOLI SALERNO

CATANZARO

POTENZA

BARI

TARANTO

CAGLIARI

SASSARI

FOGGIA

PALERMOMESSINA

REGGIO C.

d = 4.4



Restoration/Protection Schemes ComparedRestoration/Protection Schemes Compared

• RestorationRestoration– Span-Restorable Spare Capacity Assignment

(SCASCA)– Span-Restorable Joint Capacity Assignment

(JCAJCA)– Meta-Mesh (M-MM-M)

– Path-Restorable Spare Capacity Assignment (PathPath)

• ProtectionProtection– Non-Shared Backup Path Protection

(1+1 APS1+1 APS)– Shared Backup Path Protection (SBPPSBPP)



1+1 APS1+1 APS

• shortest path working route and shortest disjoint backup path (if it exists)

• at least 100% redundancy– no sharing of protection

capacity

• no optimization required– global solution is sum of

individual 1 +1 routing problems

Select(tail endtransfer)

2 diverse paths, full signal feed2 diverse paths, full signal feed

2 diverse paths, full signal feed2 diverse paths, full signal feed

Select(tail endtransfer)

2 12 2

1

SpareNeeded



Span-Restorable Schemes (Span-Restorable Schemes (SCASCA and and JCAJCA))

SCASCA• working paths first routed

via shortest path

JCAJCA• working paths optimized

jointly with spare capacity for minimum total capacity

• conceptual mesh counter-part to BLSR multi-ring networks

• can use “self-healing” distributed protocol or centralized control for KSP-type re-routing

• amenable to distributed pre-planning for fast restoration

W = 4

W = 6 multiple restoration

paths

2

4

1

3

3

13

4reuse of sparecapacity



Meta-MeshMeta-Mesh

add/dropadd/drop

add/dropadd/drop

add/dropadd/drop

expressbypass

localflow

Meta-Mesh (M-M)Meta-Mesh (M-M)

• span restoration method with logically bypassed chain subnetworks– span restoration operates on

the meta-mesh abstraction– chain subnetworks use line

loop-back for intra-chain flows

– express flows on chains fail back to meta-mesh nodes

• as JCA span-restorable capacity design but with added logical bypass spans

• especially targeted to improve mesh efficiency on sparse transport graphs

restorationflow into

meta-mesh

restorationflow into

meta-mesh

loopback

loopback

failback



Shared Backup Path Shared Backup Path ProtectionProtection (SBPP) (SBPP)

• as 1+1 APS but with sharing of spare capacity

• motivated by IETF deliberations for MPλS, etc.

• optimization chooses shared backup routes for minimum total spare capacity

• a compromise over true path-restoration to avoid signaling for stub-release

1 11 1

1

SpareNeeded

1+1 APS: 8 spare1+1 APS: 8 spareSBPP: 5 spareSBPP: 5 spare

Working path

Working path

spare capacity spare capacity reusereuse



Dynamic Path Dynamic Path RestorationRestoration

• theoretically most capacity efficient class of scheme (MCMF-like recovery)

• failure-specific re-routing of all affected demand pairs with no pre-planned disjoint backup routes

• stub-releasestub-release: surviving stubs of failed working paths released for re-use as spare capacity

• centralized control or self-organizing path restoration protocol options

• distributed pre-planning an option for very fast restoration

3 affected end-node pairs3 affected end-node pairsstub-releasestub-releaseMCMF-like recoveryMCMF-like recovery



d = 3.29d = 3.29

Making Nodal Degree a Study Parameter:Making Nodal Degree a Study Parameter:Using Families of Test NetworksUsing Families of Test Networks

• family of 18 related networks of varying average nodal degree– derived by reduction from 32-node 51-span master

network– each successively sparser network created by a random

span removal subject to retaining bi-connectivity

d = 3.23d = 3.23d = 3.16d = 3.16d = 3.10d = 3.10d = 3.03d = 3.03d = 2.97d = 2.97d = 2.90d = 2.90d = 2.84d = 2.84d = 2.77d = 2.77d = 2.71d = 2.71d = 2.65d = 2.65d = 2.58d = 2.58d = 2.52d = 2.52d = 2.45d = 2.45d = 2.39d = 2.39d = 2.32d = 2.32d = 2.26d = 2.26d = 2.19d = 2.19



Computational AspectsComputational Aspects

• full-mesh pattern of demand pairs for master network– generated by mutual attraction model (no inverse distance

effects)– average of 5.8 demands units per O-D pair

• eligible route enumeration– SCA, Path: 20 distinct eligible restoration routes per span

failure– JCA: as SCA plus 10 distinct eligible working routes per O-D

pair– SBPP: 5 distinct eligible restoration routes per O-D pair– Meta-Mesh: 20+20, 10+10

• design solutions– implemented in AMPL and solved with Parallel CPLEX 7.1 MIP– SBPP: solved within 1% of optimality (CPLEX mipgap 0.01)– All Others: solved within 0.01% of optimality (mipgap

0.0001)



500 000

600 000

700 000

800 000

900 000

1 000 000

1 100 000

1 200 000

1 300 000

1 400 000

1 500 000

2.0 2.2 2.4 2.6 2.8 3.0 3.2 3.4

Tota

l C

apac

ity

(dis

tan

ce-w

eig

hte

d)

SCA

JCA

M-M

Path

SBPP

1+1 APS

Network Average Nodal Degree, d

Results: Total Capacity vs. Nodal DegreeResults: Total Capacity vs. Nodal Degree

x2x230%30%



100 000

200 000

300 000

400 000

500 000

600 000

700 000

800 000

900 000

2.0 2.2 2.4 2.6 2.8 3.0 3.2 3.4

Wo

rkin

g a

nd

Sp

are

Cap

acit

y (d

ista

nce

-wei

gh

ted

)

SCASpare

JCA Spare

M-M Spare

PathSpare

SBPPSpare

M-M Working

JCA Working

Shortest Path Working


Results: Capacity Breakdown vs. Nodal DegreeResults: Capacity Breakdown vs. Nodal Degree

x3x3



40%

60%

80%

100%

120%

140%

160%

180%

200%

220%

2.0 2.2 2.4 2.6 2.8 3.0 3.2 3.4

Dis

tan

ce-C

apac

ity

Red

un

dan

cy

SCA

JCA

M-M

Path

SBPP

1+1 APS


40%

60%

80%

100%

120%

140%

160%

180%

200%

220%

2.0 2.2 2.4 2.6 2.8 3.0 3.2 3.4

Dis

tan

ce-C

apac

ity

Red

un

dan

cy

SCA

JCA

M-M

Path

SBPP

1+1 APS


1 ( -1)d

Results: Redundancy vs. Nodal DegreeResults: Redundancy vs. Nodal Degree

x2x2



Interpretations and Summary (1)Interpretations and Summary (1)

• capacity differences between mesh schemes come essentially all from spare capacity difference, not working

• tends to confirm that when going from ring to mesh, benefit is obtained simply by the change to mesh, regardless of type (working routing benefits greatly)

• dynamic path restoration with stub-release outperforms all other schemes in capacity efficiency

• meta-mesh and SBPP are almost as efficient as path restoration but simpler to implement



Interpretations and Summary (2)Interpretations and Summary (2)

• 1/(d-1) redundancy bound explains how span-restorable schemes react to graph connectivity (path curves are steeper)

• meta-mesh uses a span-restoration mechanism but nonetheless does better than 1/(d-1) bound

• there exists a point in the graph connectivity scale where capacity requirements level out (2.6 for this network)– helpful from a network topology planning point of view



Further ReadingFurther Reading

• J. Doucette, W. D. Grover, “Comparison of Mesh Protection and Restoration Schemes and the Dependency on Graph Connectivity,” Proc. 3rd International Workshop on Design of Reliable Communication Networks (DRCN 2001), Budapest, Hungary, pp. 121-128, October 2001.

• W. D. Grover, J. Doucette, M. Clouqueur, D. Leung, D. Stamatelakis, “New Options and Insights for Survivable Transport Networks,” IEEE Communications Magazine, vol. 40, no. 1, in press, January 2002.

• W. D. Grover, J. Doucette, “Design of a Meta-Mesh of Chain Sub-Networks: Enhancing the Attractiveness of Mesh-Restorable WDM Networking on Low Connectivity Graphs,” IEEE Journal on Selected Areas in Communications, Special Issue on WDM-based Network Architectures, in press, 1st Quarter 2002.

• W. D. Grover, J. Doucette, “Topological design of span-restorable mesh transport networks,” Annals of Operations Research, Special Issue on Topological Design of Telecommunication Networks, in press, 2001.

• W. D. Grover, J. Doucette, “A Novel Heuristic for Topology Planning and Evolution of Optical Mesh Networks,” Proc. IEEE Global Telecommunications Conference (GlobeCom 2001), San Antonio, TX, in press, November 2001.

• M. Herzberg, S. J. Bye, A. Utano, “The hop-limit approach for spare-capacity assignment in survivable networks,” IEEE/ACM Transactions on Networking, vol. 3, no. 6, pp. 775-784, December 1995.

• B. Van Caenegem, W. Van Parys, F. De Turck, P. M. Demeester, “Dimensioning of Survivable WDM Networks,” IEEE Journal on Selected Areas in Communications, vol. 16, no. 7, pp. 1146-1157, September 1998.

• R. R. Iraschko, W. D. Grover, “A highly efficient path-restoration protocol for management of optical network transport integrity,” IEEE Journal on Selected Areas in Communications, vol. 18, no. 5, pp. 779-793, May 2000.

• R. R. Iraschko, M. H. MacGregor, 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, pp. 325-336, June 1998.

• W. D. Grover, “Self-organizing Broad-band Transport Networks,” Proceedings of the IEEE, vol. 85, no. 10, pp. 1582-1611, October 1997.

• Y. Xiong; L. G. Mason, “Restoration strategies and spare capacity requirements in self-healing ATM networks," IEEE/ACM Transactions on Networking, vol. 7, no. 1, pp. 98-110, February 1999.

Overview and Evaluation of Mesh Overview and Evaluation of Mesh Network Restoration and Protection Network Restoration and Protection

Schemes Under Varying Graph Schemes Under Varying Graph ConnectivityConnectivity

John DoucetteJohn Doucette

Wayne D. GroverWayne D. GroverTRLabs and University of Alberta

Edmonton, Canada

TRLabs Smartboard Presentation01/November/2001



Span-Restorable Bound on RedundancySpan-Restorable Bound on Redundancy

W1

W2

Wd

W3

...

W1

W2

Wd

W3

W1

S2

Sd

S3

An isolated node:An isolated node:

Best case for efficiency:Best case for efficiency:

WW11 = W = Wii for all for all ii

&&SS11 = S = Sii for all for all ii



SBPP Routing InfeasibilitiesSBPP Routing Infeasibilities

• all node-pairs in (A, B, C, D) with (S, T, U, V) are unable to find a disjoint backup path if use shortest path work routing

• frequent in sparse networks

• several solutions– Suurballe’s algorithm– shortest cycle– iteration– joint capacity design

A DCB

U

V

TS

Documents

Capacity Comparison of Mesh Network Restoration and Protection Schemes Under Varying Graph Connectivity John Doucette Wayne D. Grover TRLabs and University