With the digital transformation of the communication industry and cloud-based transformation of carriers' networks, the technology-driven mode needs to switch to the business-driven mode to cope with the uncertainties of service development, business models, and technical standards. Huawei has launched a series of all-cloud solutions, among which CloudBackbone and CloudMetro are cloud-based IP WAN solutions. The IP WAN is faced with the following problems:

• E2E services are slow to deploy and change.
• Network usage is low, and network traffic cannot be adjusted globally.
• Network maintenance is complex, and fault diagnosis takes a long time.

Huawei's Agile Controller-WAN enables IP WAN cloudification and provides agile, intelligent, visible, and open capabilities for the CloudBackbone and CloudMetro solutions.

IP/MPLS Connection Automation

The Agile Controller-WAN abstracts forwarders and services, provides unified and standard network-level service interfaces northbound, and implements self-help provisioning and on-demand adjustment of VPN services. Services automatically drive tunnel creation and adjustment, greatly shortening the service rollout period and improving service provisioning efficiency. In addition, services are imported to VPNs with different QoS priorities based on tenants to provide differentiated service quality assurance.

Resource Visualization

Network-wide resources are layered and visible, and multi-layer views of physical topology, logical topology, tunnel, and VPN services are provided, facilitating management and planning, automatic discovery of network resources, improving the O&M efficiency, and shortening the network initialization time. Real-time monitoring of service SLA information, such as bandwidth, status, latency, packet loss, jitter, and service connectivity, changes passive O&M to proactive O&M. Intelligent traffic optimization capabilities allow service traffic to be quickly adjusted and network congestion to be quickly relieved, improving the operation efficiency by several times.

Network Optimization

The Agile Controller-WAN provides an intelligent network optimization solution for IP and MPLS networks to implement real-time automatic traffic optimization and SLA information collection and display.

MPLS Network Optimization

Based on the Routing Optimization Algorithm based on Matrix (ROAM) algorithm, the MPLS network is capable of multi-dimension optimization in bandwidth, latency, cost, and priority. End-to-end optimal paths are computed for MPLS TE tunnels network wide to ensure an optimal path for a single service and balance of global network traffic, which effectively improves bandwidth utilization and minimizes network expansion costs.

IP Network Optimization

Topology management, path calculation, and outgoing traffic optimization allow for network adjustment in the IP backbone domain and real-time response to traffic changes. In addition, users can learn traffic distribution of the entire network and plan traffic through the graphical user interface (GUI), which provides one-hop scheduling capabilities for cross-domain outgoing traffic on a native IP network. This also implements flexible bandwidth for VIP traffic, intra-city traffic localization, and egress link balancing as well as providing one-hop or multi-hop optimization capabilities within a domain, improving backbone network resource utilization and operation efficiency.

Open Northbound Interfaces

RESTCONF or RESTful is the main northbound protocol. The Agile Controller-WAN uses RESTful or RESTCONF to connect to orchestration-layer devices and third-party applications, and to receive service information and policies from those devices and applications. The standard API open network capability and standard YANG model are used to accelerate service innovation. The controller supports network service programmability and interconnection with third-party orchestrators or apps. With open northbound interfaces, the controller helps customers smoothly evolve their networks, build an open SDN ecosystem, and achieve business success faster.

High Availability (HA)

The controller provides HA, which involves local cluster and geographical redundancy. Cluster nodes process services independently. If a single node is faulty, services can be automatically switched to other nodes. If more than half of the nodes are normal, the controller cluster can provide services properly. Carrier-class reliability is provided, and data consistency is supported. If configuration inconsistency occurs, configuration reconciliation is performed to detect and rectify differences. In this way, configurations on the southbound and northbound interfaces are consistent.