1. Flexible Information model
LSO must hold an inventory of the services that it is orchestrating, and it must associate each service to the resources used to provide that service, whether they are implemented in a physical network element, in a VNF, or are delivered by a third party provider. Since every service provider has unique service definitions, as well as unique network topologies, policies and technologies, the model must be able to be rapidly customized in each deployment.
2. Rapid Data Source Integration
This information model must be populated with the set of services to be orchestrated. The ability to instantiate a new service only happens once in the lifecycle of each service whereas service changes and assurance need to be addressed on an ongoing basis. The latter functions cannot be carried out until the service inventory is populated and the data sources required to do this can be very extensive. For example, configuration files from physical network elements will generally be one trusted data source, but to build the complete service inventory, there is much information that the network elements do not know. As a result, a design to allow rapid adaptation and continuous ingest of these data sources is required.
3. Data AuditIn most every existing network, different data sources will not agree on particular information about the service. The LSO must be able to analyze and present these mismatches to users in order to ensure that the service information model is accurately populated.
4. Service Visualization
Different management views of the transport network need to be supported, including fulfillment, performance management, troubleshooting, network optimization and just in time capacity management views.
5. Service Assurance & Analytics
Typically, operators need to manage a very large number of existing services and the assurance phase is the longest part of their lifecycle. The LSO must be able to take in a tremendous amount of real-time data (such as fault, utilization, service OAM, RAN statistics, NFVI compute information) to support rapid troubleshooting and repair as well as for predictive analytics and capacity planning purposes. The LSO needs to turn all this data into meaningful information so that it can identify issues with the largest service impacts and then take corrective actions that will result in the best quality of service for the customers.
6. Service Workflow
Translating a service request into a completed action typically involves multiple steps and a state machine. In many cases (for example, when a physical UNI needs to be installed), human activity is required. Some requests may require execution during maintenance windows and may need to be monitored. Sometimes, the request involves placing orders to third party access vendors. Ideally, the vast majority of the steps in each workflow are fully automated, and in some cases (for example, a bandwidth increment where underlying network capacity is sufficient), the entire workflow can be automatic. An LSO needs to be able to handle workflows that are both automated and which require human intervention. It should also allow workflows to be customized to the particular service provider requirements.
7. Interaction with MANO
One of the steps in many service workflows will be to instantiate or change the capacity of VNFs, such as virtual firewalls and other Layer 4-7 functions. The LSO will need to trigger this request and ensure it is successfully executed.
One of the steps in a service workflow will be to provision the initial creation of the service, as well as the many changes through the service lifecycle (for example, add an endpoint, change firewall rules and eventually disconnect the service).
9. Coordinated Testing
Another step in the workflow is to set up and execute testing to ensure that service instantiation or changes have been completed successfully.
10. Off-net Ordering
In many cases, connectivity services require third party service providers to reach all required locations. In this case, fulfilling service requests can mean triggering orders to third parties and tracking them to fulfillment.
For more information on the requirements for lifecycle service orchestration download CENX and Heavy Reading’s white paper, “NFV’s Biggest Barriers: Overcoming the Operations Challenge”.