NOC’2003 Vienna 1
Resilience with Tailored
Recovery Time in Switched Optical Networks
T. Jakab, Zs. Lakatos, jakab@hit.bme.hu
Department of Telecommunications,
Budapest University of Technology and Economics Budapest, Hungary
NOC’2003 Vienna 2
Outline
• Motivations
• Optical Channel Based Services in Switched Intelligent Optical Networks
• Potential Resilience Options for Permanent Optical Channel Based Services
• Resilience with Tailored Recovery Time
• Summary and Conclusions
NOC’2003 Vienna 3
Motivations
• Intelligent flexibility is required in optical networks
– to cope with traffic uncertainties,
– to enable fast optical channel provisioning, – to support complex shared capacity based
resilience.
• Efficient resilience schemes are of increased importance in optical networks carrying highly concentrated traffic.
• Service differentiation is a crucial point to
support different client services and improve service profitability.
NOC’2003 Vienna 4
Optical Channel Based Services
• Transport Services
– Permanent optical channel service,
– Soft-permanent optical channel service, – Lambda trunking service,
– OVPN service.
• Service Requirements – Fast provisioning,
– Differentiated services,
– Enhanced service resilience.
NOC’2003 Vienna 5
Network Resilience
• Dynamic application of extra network resources to limit the impact of failures
• Based on dedicated or shared network resources
• Requires intelligent switching function to be supported in nodes
• Basic schemes:
– 1+1 dedicated protection, – n:m shared protection,
– restoration (failure state dependent dynamic configuration of shared capacities).
NOC’2003 Vienna 6
Illustrative numerical results
• Analysis of different resilience options – Full flexibility implies restoration
• Resilience of switched OCh based services
– Service oriented considerations – restoration with tailored recovery time
NOC’2003 Vienna 7
Illustrative examples (1)
Resilience in Full Flexible Networks
0%
50%
100%
150%
200%
250%
300%
not prot.
optimal path rest.
1+1 Resilience Cases
Relative Hop*OCh
Extra for resilience Working
0%
50%
100%
150%
200%
250%
300%
not prot.
full flex.
1+1 term.
switch.
1+1 full flex.
optimal path rest.
full flex.
Resilience Cases
Relative #Switch Ports
Extra due to resilience Working
• Full flexible network: each capacity unit terminates on switching-capable node equipment (e.g. flex. OADM or OXC)
Comparison of link capacities Comparison of switch capacities
Extra for dedicated resilience
Savings on capacity
sharing Switches
not used for resilience
NOC’2003 Vienna 8
Illustrative examples (1a - Link capacities)
Resilience in Full Flexible Networks
0%
50%
100%
150%
200%
250%
300%
not prot.
optimal path rest.
1+1 Resilience Cases
Relative Hop*OCh
Extra for resilience Working
0%
50%
100%
150%
200%
250%
300%
not prot.
full flex.
1+1 term.
switch.
1+1 full flex.
optimal path rest.
full flex.
Resilience Cases
Relative #Switch Ports
Extra due to resilience Working
• Full flexible network: each capacity unit terminates on switching-capable node equipment (e.g. flex. OADM or OXC)
Comparison of link capacities Comparison of switch capacities
Extra for dedicated resilience
Savings on capacity
sharing Switches
not used for resilience
NOC’2003 Vienna 9
Illustrative examples (1a)
Link capacities for Resilience in Full Flexible Networks
• Full flexible network: each capacity unit terminates on
switching-capable node equipment (e.g. flex. OADM or OXC) Comparison of link capacities
0%
50%
100%
150%
200%
250%
300%
not prot.
optimal path rest.
1+1
Resilience Cases
Relative Hop*OCh
Extra for resilience
Working Savings on
capacity sharing Extra for
dedicated resilience
NOC’2003 Vienna 10
Illustrative examples (1b - Switch capacities)
Resilience in Full Flexible Networks
0%
50%
100%
150%
200%
250%
300%
not prot.
optimal path rest.
1+1 Resilience Cases
Relative Hop*OCh
Extra for resilience Working
0%
50%
100%
150%
200%
250%
300%
not prot.
full flex.
1+1 term.
switch.
1+1 full flex.
optimal path rest.
full flex.
Resilience Cases
Relative #Switch Ports
Extra due to resilience Working
• Full flexible network: each capacity unit terminates on switching-capable node equipment (e.g. flex. OADM or OXC)
Comparison of link capacities Comparison of switch capacities
Extra for dedicated resilience
Savings on capacity
sharing Switches
not used for resilience
NOC’2003 Vienna 11
Illustrative examples (1b)
Switch capacities for Resilience in Full Flexible Networks
• Full flexible network: each capacity unit terminates on
switching-capable node equipment (e.g. flex. OADM or OXC)
0%
50%
100%
150%
200%
250%
300%
not prot.
full flex.
1+1 term.
switch.
1+1 full flex.
optimal path rest.
full flex.
Resilience Cases
Relative #Switch Ports
Extra due to resilience Working
Comparison of switch capacitiesSwitches not used for resilience purposes
NOC’2003 Vienna 12
Network Resilience Service Considerations
• Different applications may need resilience with different characteristics, such as
– recovery speed, or
– rate of recovered capacity (partial/entire).
• Different resilient classes can be specified according to the different needs
• Aim: meet different resilience requirements on the same technical basis and lowest cost
NOC’2003 Vienna 13
Illustrative examples (2)
Restoration with Tailored Recovery Time
• Recovery time is assumed to be proportional with the
number of active switching nodes involved in the process, and with the processing load of each active switch
• Some switches can be pre-set and fixed to speed up the recovery process
• Reduced flexibility results in less efficient capacity sharing, therefore the amount of extra resources for restoration is increasing
• The joint optimisation of different classes may reduce the penalty
NOC’2003 Vienna 14
Illustrative examples (2)
Restoration with Tailored Recovery Time
Working path 1Working path 2 Single
Failure 1
Single Failure 2
Recovery path 2 Recovery path 1
NOC’2003 Vienna 15
Illustrative examples (2)
Restoration with Tailored Recovery Time
Working path 1Working path 2 Single
Failure 1
Single Failure 2
Recovery path 2 Recovery path 1 Pre-set
switches
Capacity sharing
NOC’2003 Vienna 16
Illustrative examples (2)
Restoration with Tailored Recovery Time
Working path 1Working path 2 Single
Failure 1
Single Failure 2
Recovery path 2 Recovery path 1 Pre-set
switches
No capacity sharing
NOC’2003 Vienna 17
Illustrative examples (2)
Restoration with Tailored Recovery Time
Traditional restoration 1+1
dedicated protection
Resource needs for networks with different resilience options
0%
20%
40%
60%
80%
100%
120%
1+1 path prot.
single hop path rest.
min. path rest.
double hop path
rest.
double and triple
hop path rest.
capac.
opt. path rest.
Resilience options
OCh*hop
Recovery path statistics for different resilience options
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50
1+1 path prot.
single hop path rest.
min. path rest.
double hop path
rest.
double and triple
hop path rest.
capac.
opt. path rest.
Resilience options Average logical hop count
spare for resilience working
NOC’2003 Vienna 18
Illustrative examples (2a - Average Logical Hop Count)
Restoration with Tailored Recovery Time
Resource needs for networks with different resilience options
0%
20%
40%
60%
80%
100%
120%
1+1 path prot.
single hop path rest.
min. path rest.
double hop path
rest.
double and triple
hop path rest.
capac.
opt. path rest.
Resilience options
OCh*hop
Recovery path statistics for different resilience options
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50
1+1 path prot.
single hop path rest.
min. path rest.
double hop path
rest.
double and triple
hop path rest.
capac.
opt. path rest.
Resilience options Average logical hop count
spare for resilience working
NOC’2003 Vienna 19 Recovery path statistics for different resilience
options
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50
1+1 path prot.
single hop path rest.
min. path rest.
double hop path
rest.
double and triple
hop path rest.
capac.
opt. path rest.
Resilience options Average logical hop count
Illustrative examples (2a)
Restoration with Tailored Recovery Time
Average Logical Hop Count
1+1 dedicated protection
Traditional restoration Single logical
hop, receiver end switching only
Fastest
Multiple logical hops, switching in each node via
the path
Slowest
NOC’2003 Vienna 20
Illustrative examples (2b - Resource Needs)
Restoration with Tailored Recovery Time
Resource needs for networks with different resilience options
0%
20%
40%
60%
80%
100%
120%
1+1 path prot.
single hop path rest.
min. path rest.
double hop path
rest.
double and triple
hop path rest.
capac.
opt. path rest.
Resilience options
OCh*hop
Recovery path statistics for different resilience options
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50
1+1 path prot.
single hop path rest.
min. path rest.
double hop path
rest.
double and triple
hop path rest.
capac.
opt. path rest.
Resilience options Average logical hop count
spare for resilience working
NOC’2003 Vienna 21 Resource needs for networks with different
resilience options
0%
20%
40%
60%
80%
100%
120%
1+1 path prot.
single hop path rest.
min. path rest.
double hop path
rest.
double and triple
hop path rest.
capac.
opt. path rest.
Resilience options
OCh*hop
Illustrative examples (2b)
Restoration with Tailored Recovery Time
Resource Needs
Highest extra for resilience
(No sharing - 1+1 dedicated protection
only)
Capacity optimal restoration without
recovery time specifications
Same routing without effective capacity constraints
(simplified case)
From faster to slower
spare for resilience working
Demand classes with different recovery time
requirements,
different classifications
NOC’2003 Vienna 22
Summary on Restoration with Tailored Recovery Time
• Different clients and applications require different recovery times
• Applying restoration, to shorten the the recovery time some switches can be pre-set via the restoration paths
• Pre-set switches decrease de resilience capacity sharing efficiency, therefore the resilience related extra capacity increases
• Joint optimisation of multiple recovery time classes decreases this penalty
NOC’2003 Vienna 23
Conclusions
• Intelligent switching introduced in optical networks enables enhanced resilience schemes
• Shared capacity oriented restoration is the cost effective solution for the resilience in full flexible networks
• Demand classes of different recovery times enables service differentiation, however the joint optimisation of these classes results in low capacity penalty