A Cloud native Network Function (CNF) is software which replaces specialized hardware, operating in the cloud.
Examples of CNFs can include network routers, bridges, firewalls, and Virtual Private Network (VPN) gateway services. Cloud native Network Functions replace Physical Network Functions (PNFs). PNFs are purpose-built hardware devices (with embedded software or systems) which provide essential network services.
CNFs are conceptually similar to Virtual Network Functions (VNFs), a fundamental component of a Network Functions Virtualization Infrastructure (NFVI). Where they differ is implementation: “Virtual” functions imply the presence of resource-intensive virtual machine operations.
“In a VM, virtualization abstracts the hardware and runs multiple OS instances,” explains Intel. “In contrast, containers abstract the OS, multiple containers on a server share a common OS kernel. In addition to containers, CNFs use orchestration services that automate service delivery.”
What cloud native means
CNFs are built using a microservices architecture, operating natively in the cloud. Cloud Native Network Functions replace the monolithic application model of computing with individual, interconnected microservices. These microservices are housed in software containers. Containers are run-time environments which include all the application code, binaries, and dependencies needed for the microservices to operate.
CNFs share common characteristics besides containerization and orchestration. They can be spun up and deactivated almost instantly. Cloud native network functions support the dynamic scale and elasticity required to operate at hyperscale levels. They operate using fewer compute resources than virtual functions. Orchestration systems like Kubernetes enable such services to operate at datacenter scale and hyperscale.
Following agile DevOps principles, cloud native network functions can be optimized, improved, and enhanced continuously throughout their lifecycle. Cloud native network functions are reusable and flexible. They act as common infrastructure that enables the rapid creation and deployment of new apps and services over time.
“Cloud native technologies empower organizations to build and run scalable applications in modern, dynamic environments such as public, private, and hybrid clouds. Containers, service meshes, microservices, immutable infrastructure, and declarative APIs exemplify this approach,” said the Cloud Native Computing Foundation (CNCF).
The CNCF, in concert with Linux Foundation Networking (LFN) has provided an open-source Cloud native Network Function Testbed. The CNF Testbed provides a baseline to test network functions as both VNFs and CNFs to compare performance.
“As telecom architecture evolves from VNFs into CNFs, the benefits include cost savings from improved bin packing, higher development velocity, and resiliency to failures of individual CNFs, machines, and even data centers,” said the Linux Foundation.
Cloud Native Network Functions and 5G
The 3GPP’s 5G Core Network standard employs a Service Based Architecture (SBA). This cloud-native approach to the core network enables and ultimately requires the disaggregation of network hardware and software. This is needed to meet the scale, performance, and varying service demands of 5G enterprise and users.
“Cloud-native principles and technology have proven to be an effective acceleration technology in building and continuously operating the largest clouds in the world,” said Cisco.
A cloud-native 5G network provides the completely digitized platform necessary to deploy new cloud services and to take advantage of cloud-native 5G benefits like massive IoT, edge computing, and network slicing.
Cloud Native Network Functions will ultimately help operators pivot from Non-standlone (NSA) 5G architecture which depends on a 4G core network to operate, to standalone (SA) 5G. Standalone 5G pairs 5G radios with a cloud-native 5G core network.
“Other benefits that a cloud-native architecture based on microservices brings to 5G included stateless and stateful services, a common and shared data layer, abstracted infrastructure and streamed telemetry data,” said Intel. “But to achieve these benefits, there is one simple truth: traditional, relative slow IP routing-like networking in the data center cannot cope with a cloud-native system that supports service discovery and that creates, moves and stops microservice instances in real time. Therefore, CoSPs must learn the basic design principles for building a cloud-native architecture.”