This evolution means moving away from proprietary hardware to a programmable software-defined network delivering a dynamic environment which can respond in real-time to shifts in services. Enabling the vast array of new 5G applications, for example, can only be realized when operators evolve their networks to 5G SA (Stand Alone), a cloud-native 5G Core (5GC) based on NFV built using IT network principles.
Your network is your powerful data and analytics engine. You have a mass of data running through the network and this data can be turned into real intelligence and insight not only supporting your network priorities but also your business priorities to grow the extraordinary opportunities 5G enables. Subscriber-generated data is your network gold to harness for not only your own applications and analysis but also for partner and external applications. By leveraging the network intelligence, you can address Revenue, Service and Saving Money, ensuring the network objectives are fully aligned with the business objectives.
When building or expanding the network by leveraging a virtualized infrastructure, the visibility engine that will provide Service Assurance and Analytics can no longer be addressed during the later phases – it must be designed in at day one with essentially the same infrastructure being used for the Assurance system itself. By ensuring the systems are integrated, you can be confident you have seamless visibility across the multiple technologies (2G, 3G, 4G, 5G) as well as across RAN and Core.
Assurance Challenges with Virtualization and NFV
Any Service Assurance and Analytics system must adapt to the more open and flexible domain virtualization brings.
- Traffic Visibility
A key element of any Assurance system is getting access to the underlying data and transforming this into intelligence and insight that can be used across the organization. In a static environment once the capture engines are deployed there is often very little change to contend with. In a Virtual environment that is highly dynamically changing requires a more flexible methodology to adapt to changes in real-time, integrated into the management and orchestration (MANO) systems. NFV-based services will be dynamic and involve many different VNFs with network topologies adapting and breathing.
Any shared resource environment requires contention management enabling the smooth operation of critical tasks. Any shared resource – compute power, memory, I/O, processes can have an impact. Although container-based architectures are inherently light-weight they are not resource free. In addition, changes in traffic patterns and growth can significantly affect performance.
- Continuous Change
As previously mentioned, the dynamic nature of NFV-based services supports an ever-changing network topology model which in turn requires a more flexible and adaptable service rollout system. The use of a cloud-native architecture enables the capability of CI/CD (Continuous integration and Continuous Delivery & Deployment) bringing together development and operations in an agile framework. Sub-components can be tested and deployed quickly underpinning a significantly more flexible, adaptable, and programmable network. Not only does this improve the day to day Assurance and Optimization of the network, it also delivers analytics, enabling data science and business automation.
- Ecosystem Integration
In an NFV environment, operators need a set of integrated solutions from many vendors to deliver the network and service platform. Standards continue to evolve and there is continuous learning and innovation particularly now as operators embrace 5G SA (Stand-Alone). It is clearly vital that operators can rely on the more complex, multi-vendor, open architecture to continue to deliver benefits of the 5GC as the network grows and evolves.
Next Generation Assurance
Any Assurance architecture for a Virtualized environment must adopt certain characteristics and capabilities to thrive in the new 5GC enabling seamless network, service, and customer visibility. As operators embrace 5GSA and a Virtualized Core Network along with Open RANs, VIAVI continues to work closely with our customers, technology partners and industry standard groups to virtualize the NITRO Mobile portfolio of Core and RAN Assurance and Analytics solutions. The acquisition of Expandium cements VIAVI’s leadership position and accelerates our ability to deliver innovative 5G solutions to support even the most aggressive virtualization initiatives.
- Cloud-native, Microservices-based
The NITRO Mobile microservices-based architectural approach arranges an application as a collection of loosely coupled services lending itself to continuous delivery. Individual microservices can be individually scaled, pipelined across VNFs (Virtual Network Functions) and deployed either regionally or centrally often in a containerized environment. This provides flexible and significant horizontal scaling together with “open” records for orchestration and control. The cloud-native approach means applications can run in an elastic and distributed nature in a public, private or hybrid cloud.
- Stream Processing
A fundamental element of the NITRO Mobile virtualized platform architecture is the capability of performing real-time stream processing. This allows xDRs from probes, agents, NEM measurements and other external sources to be sent to the Data Message Bus and then processed according to the set of rules such that KPIs, correlated xDRs and xVRs (Customer Value Records) are generated immediately. This flexibility delivers the performance required for massive scale while allowing the intelligence to be parametrized around Transactions, Services and Customers. In addition, parallel stream processing delivers greater efficiency with less database storage requirements than traditional methods resulting in an improved Total Cost of Ownership (TCO).
- Data Capture
Capturing the data in the many points of the network underpins any Assurance system. There are various methods that can be employed depending on the operator’s architectural approach. In a VNF environment, packet access can be either physical or virtual. Many Virtual User Plane Network Elements are using SR IOV (Single Root Input/Output Virtualization) to provide higher data rates. With SR IOV, a Virtual Machine can gain shared direct access to physical network cards installed in the hypervisor.
Using SR IOV physical tapping can be deployed to the fibers entering and leaving the element or by taking a span from the “Top of Rack” switch. This by its nature would not provide any visibility into East-West traffic exchanged inside the element. For a Virtual Network today, it is possible to physically tap the Fibers at the Gateways and get visibility into all the traffic inside the Virtual Evolved Packet Core.
With a pure virtual tapping approach, virtual agents are deployed inside the VNF and packers are mirrored/tunneled typically using GRE (Generic Routing Encapsulation) or VXLAN encapsulation. Virtual tapping continually evolves. It can for example be based on Open vSwitch (OVS), or Kubernetes Container Network Infrastructure (CNI).
If a Service Communication Proxy (SCP) system is deployed, then all communication from a trusted zone to the outside world goes through them. It is possible to mirror the packets out of the SCP.
Hybrid Virtual Tapping
A Hybrid approach may employ physical tapping and simply mirror that data from the agents, embracing both physical and virtual methods.
SmartNICs, as a developing technology can be used to accelerate networking functions in hardware, reducing requirements on the Hypervisor node. They are intended to provide the performance, functionality and programmability for compute nodes and bare metal servers.
As discussed above, SR IOV provides a methodology within the bus itself where multiple Virtual Machines can share the hardware resources. Any VM can talk directly to for example an Ethernet controller
Data can be gathered by capturing events from the specific Network Elements in question. This could be where the capture engine subscribes to or polls for new events or often both depending on the NE Manufacturer
Note that often a NITRO Mobile system is employing many of these methodologies for data capture to gather the data required including polling and streaming of information across various physical and virtual data points.
With an NFV domain comes the need for a more flexible, dynamic, and reactive Management and Orchestration (MANO) framework. In particular, the orchestrator is responsible for firing new network services and VNF packages and as a result should be tightly integrated with any Services Assurance system deployed. With the NITRO Common Infrastructure (NCI), intra-VNF management is supported through self-monitoring of each VNF delivering telemetry to a central location. Any install, update, and scale out/in would be communicated. Inter-VNF orchestration is also supported to an external orchestrator through methods such as ONAP and OSM. VIAVI can also provide the localized orchestrator deployed as a domain controller and feeding to a master orchestrator.
ONAP (Open Network Automation Platform)
ONAP is an Open Source initiative focusing on network automation, delivering a platform for orchestration and management. The ONAP architecture enables VNFs to be interoperable in an automated, real-time environment.
Automation and Orchestration now lend themselves to a cloud-based architecture where functions that are normally managed within the firewall of a service-provider are now being cloudified. This is a major evolution in the design and operation of a network that can bring significantly lower costs and reduced complexity as well as delivering on the flexible architecture Virtualization brings. NITRO Mobile is fully aligned with this open cloud-based approach, embracing the elements within the platform for openness, flexibility, cost efficiency.
Network Data Analytics Function (NWDAF)
The need for Analytics to support a broad range of lines of business in the operator supporting data science and revenue-generating opportunities is well understood. The Network Data Analytics Function (NWDAF) is part of a 3GPP standard for the 5GC as defined in TS 29.520. It provides standard interfaces from a service-based architecture for centralized data collection and analytics through subscription or by request. The proposal would have a distributed architecture providing data at the edge in real-time along with a central aggregator of analytics.
NWDAF is a key element in the provision of automated assurance and analytics allowing operators to manage in real-time elements such as network slicing in a cloud-native NFV domain. A central foundation of NITRO Mobile is an open platform with northbound and southbound interfaces aligning with standards and technologies as they evolve. NITRO Mobile would be able to leverage the NWDAF principles to provide granular insight on the 5GC network, services, and customers.
Automation, ML and AI
Automation takes the principles of Management, Assurance, and Optimization and encodes them in a process that can be performed automatically with minimal or even no human intervention. The complexity of 5G and Virtualization demands a level of automation allowing engineers to manage other aspects of network operations and planning. With the dynamic nature of these networks, the number of parameters involved and with often disparate verticals being supported, many would argue some level of Automation is a must. Indeed, the openness and flexibility offered through a Virtual approach in a multi-vendor environment can really only be fully realized with automation.
The concepts of Machine Learning (ML) and Artificial Intelligence (AI) may be the cornerstone of Automation. These elements have been key components of the NITRO Mobile evolutionary capabilities. Both supervised and unsupervised ML are used within the solution today along with a branch of AI called evolutionary computing. Many ML algorithms and models thrive on data in large quantities and these algorithms and models in turn underpin the AI required to automate processes.
AI has a variety of definitions, and in the context of Automation, we regard AI as using an automated system to solve a problem that humans intuitively can solve but that computers typically find hard. In contrast to AI, ML is the creation of models that describe relationships between inputs (features) and outputs (targets).
ODE, Analytics and BI
The evolution of 5G along with the complexity of virtualization demands a new approach to Assurance. There are 3 primary workflows associated with Assurance, Analytics and Business Intelligence, namely:
- Operations: Allows the identification of subscriber impacting anomalies and QoE degradation
- Diagnostics: Troubleshoot and diagnose root cause to effect timely resolution
- Exploration: Data Science and Advanced Analytics
Underpinning these workflows is the need to understand the separation of data requirements albeit from the same data source. In many ways this is similar to the concept of network slicing where a multiplex of virtualized networks on the same physical infrastructure can be applied to deliver often disparate vertical needs. With NITRO Mobile, we deliver the unique Intelligent Assurance Slice Manager (IASM) which delivers assurance slices per ODE workflow applying dynamically resources and priorities to the workflows in real-time. Flexible data processing pipelines of microservices will enable the workflows bringing a new level of automation and data science.
5G Converged Assurance Solution
Often service providers monitor the RAN and the Core performance independently usually relying on different sets of metrics backed up by different teams of experts. The converged assurance solution delivered as part of the NITRO Mobile portfolio provides the combination of RAN and Core monitoring data into a single dashboard, reports, and workflows. This means 5G RAN and Core data can be integrated, and a service-view provided across the evolving virtualized network.