Protection and restoration fundamentals
Telecommunications infrastructure spans cities, continents, and oceans. It is exposed to fiber cuts, equipment failures, loss of power, and natural events. A backhoe cutting a buried fiber, known in the industry as backhoe fade, can disconnect a city from the network. Each defect can interrupt an established transmission path and reduce service availability.
Network design therefore considers reliability and availability alongside capacity. A "five nines" availability requirement () corresponds to less than 5.26 minutes of unavailability per year. Automated protection and restoration schemes can switch traffic or establish another path after a defect is detected. Protection uses capacity assigned in advance and switches traffic to a ready resource. Restoration computes or establishes a route after the failure and can use available capacity. This distinction affects reaction time and the cost of reservation.
Reliability principle: redundancy
Protection mechanisms use redundancy, meaning an additional resource that can carry traffic after a working path defect. A typical model creates at least two distinct paths between important nodes:
- Working or primary path: The main route used for data transmission during normal operation.
- Protection or backup path: An alternative route or resource kept ready to carry traffic after a working path defect.
Shared risk of failure
Geographic and physical path disjointness improves protection against shared failures. Routing the working and protection paths through separate cables, different ducts, and, where possible, different physical routes reduces the probability of losing both resources in one incident. Placing both cables in one duct allows the same backhoe to cut both paths in one event.
Linear protection schemes
A direct link between two nodes, such as Node A and Node B, can use three common linear protection variants. They differ in resource reservation and in how additional traffic is handled.
1+1 protection
This is the simplest of the described schemes. Traffic is permanently bridged at the ingress and transmitted on both the working and protection paths. The egress monitors both signals and selects the one with better quality. After a working path defect, selecting the protection signal does not require a new connection to be established.
- Pros: Very short switchover, typically below 50 ms in common mechanisms, and simple logic.
- Cost: The protection resource continuously carries a copy of the traffic and cannot carry additional traffic. Using all redundant capacity roughly halves the network capacity available for other traffic.
1:1 protection
One protection path is assigned to one working path, but during normal operation it can carry lower-priority additional traffic. After a defect, the nodes first remove that traffic from the protection path and then switch the protected traffic from the failed working path onto the released resource. The two ends of the connection must coordinate their state during the switch because the additional traffic must be removed.
- Pros: The protection resource can carry additional traffic during normal operation.
- Cost: Additional traffic may be lost, and the switch requires signaling and coordination.
1:N protection
This is a resource-efficient linear variant. One protection resource is shared by multiple () working paths. After a defect on one of the paths, its traffic can be directed to the shared protection path if capacity and priority rules allow it.
- Pros: Sharing reduces the amount of reserved protection capacity.
- Limitation: The scheme can protect only one failure in the group at a time. If two working paths fail simultaneously, shared capacity can switch only the path allowed by the available resource and priority rules.
Ring protection mechanisms
In a ring topology, nodes are connected by two directions around a closed loop. This redundancy is used in metropolitan and regional networks. After a break in one span, traffic can be directed the long way around through the remaining part of the ring. UPSR and BLSR are protection scheme names used in SONET and SDH systems.
UPSR (Unidirectional Path Switched Ring)
UPSR is the ring equivalent of 1+1 protection. It uses two fibers or rings, a working and a protection path, with traffic flowing in the same direction. The source sends the same copy of the traffic along both routes, and the destination selects the signal with the better state. Switchover after a failure is nearly instantaneous, but the scheme continuously sends a copy and limits capacity utilization.
BLSR (Bidirectional Line Switched Ring)
BLSR uses traffic in both directions around the ring. In a common two-fiber variant, one fiber carries traffic in one direction and the other carries it in the opposite direction; half of each fiber capacity is assigned to working traffic and the other half is reserved for protection. This can use working capacity more efficiently than UPSR, but it requires coordinated switching.
The key mechanism is loop-back. After a break between nodes B and D is detected, adjacent nodes place traffic intended for the failed span on reserved protection capacity and send it around the opposite side of the ring. The ring then provides an alternate path between the endpoints of the failed span and can restore all traffic when the available capacity is sufficient.
Restoration in a mesh network
A can offer several routes around a failed element. Protection reserves resources in advance, while a mesh network has no predefined idle protection paths. Restoration selects or activates an alternate path after a defect is detected.
After a defect, a central SDN controller or management system uses a global view of network state, selects a path that meets the requirements, and remotely configures the optical switches OXCs along it. Mesh paths can be calculated after the failure or precomputed and activated when it occurs. Restoration is usually slower than protection, from hundreds of milliseconds to a few seconds, and does not provide the same time guarantee. Its benefit is that resources can be shared by many connections because capacity remains available until a failure occurs.