Model Driven Service Assurance using NFV Orchestrator
In a previous blog post, we discussed the Top-6 challenges for Service Assurance in NFV: Scale, Vendor Compatibility, PNF Integration, Lack of Instrumentation, Lack of Unified View and Field Extensibility.
How NFV Orchestrator helps with service assurance?
An Orchestrator such as Anuta NCX uses YANG models to automate network services across multi-vendor physical and virtual infrastructure. System Integrator partners can easily extend these YANG models to customize the deployment for various use-cases including branch, campus, data center and cloud.
An Orchestrator leverages abstraction to separate the service logic from the underlying device semantics. While it is not practical to validate all the combinations of VNF, NFVI and VIM components, the decoupling of layers minimizes interdependency among multiple vendor VNFs. For instance, Anuta NCX has been validated to deliver service chaining among 35 different industry leading vendors providing L2-L7 PNFs and VNFs.
As part of the YANG service model definition, network architect can define SLA parameters (e.g. which interfaces and metrics to monitor). The YANG model also includes remediation steps when the SLA parameters are not met.
During service provisioning, Network Orchestrator validates the service model against the existing infrastructure and ensures capacity prior to provisioning the network elements. The Network Orchestrator generates commands and APIs for multi-vendor infrastructure as per the recommended best practices. Further, these commands are sent in an atomic fashion, so if any one device fails to provision, Orchestrator will roll back the configs from rest of the devices.
After service is up and running, admin can schedule periodic reconciliation tasks. Network Orchestrator uses its service model definition to discover the existing configs and makes sure the policy is consistent with underlying infrastructure.
Network Orchestrator constantly collects the SLA metrics and if any parameters are violated, it executes the remediation steps to automate corrective actions as per the YANG model.
An Orchestrator such as Anuta NCX has built-in device models for multiple vendor devices and platforms including physical and virtual form factor. The same service model can be executed on physical, virtual or hybrid infrastructure. In case of NFV, the Network Orchestrator manages the complete life-cycle of VNFs including instantiation, licensing, service provisioning, auto-scaling as well as decommissioning. The Orchestrator provisions and monitors health of a service that spans multiple physical and virtual elements.
Network Orchestrator provides a single pane of glass for the entire service regardless of the infrastructure differences. For any given tenant, admin can view the service status, provisioned operations, current and historic SLA metrics as well as any alarms related to service health. This information is available through REST API for integration with OSS/BSS for billing purposes.
NFV results in many small VNFs replacing one physical appliance thus creating management challenges. The good news is that the network monitoring vendors are also delivering their products as VNFs.
An Orchestrator such as NCX can quickly spin up one VNF (e.g. HPE Arc Sight VNF) per tenant as part of the service activation.
The Orchestrator itself has a distributed server-agent architecture.
The server provides the administrator a central view to create business and service policies. The agent is deployed closer to the network infrastructure and designed to manage the network resources within the pods including policy provisioning and management. This allows Orchestrator to scale across multiple data centers providing uniform business and network policy compliance. Multiple servers can be setup in an Active-Active configuration behind a load balancer to share the load and to offer high availability.
Service Assurance in NFV is still an evolving subject. The Network Service Orchestrator such as Anuta NCX has promising architecture and features.
– Kiran Sirupa, January 6th, 2017.