The architecture of 5G Core (5GC) uses a Service-based Architecture (SBA) framework, which employs a Service-based Interface (SBI) between the identified network functions within the 5GC Control Plane. For the 5GC User Plane Functions (UPF) and Radio Access Network (RAN), the Evolved Packet Core (EPC) architecture reference points are still used. However, they include specific updates and enhancements to accommodate the requirements of 5G technology.
The following diagram depicts the service-based 5GC. The 5GC SBI is designed as a communication bus using the HTTP/2 protocol for Control Plane Network Functions in the style of a RESTful API for scaling.
Figure 6. 5G Core system-based architecture
5GS SBA and SBI support stateless network functions, where the compute resource is decoupled from the storage resource. This design enables a cloud-native microservices architecture for 5GC as standardized by the European Telecommunications Standards Institute (ETSI). The diagram depicted in Figure 6 presents the evolutionary process of the Telco transformation. It transitions towards a multicloud landscape that caters to a diverse range of network functions (NFs). These NFs exhibit distinct requirements and are delivered and packaged in varying software forms.
Figure 7. Telco cloud landscape—home of a telecom multicloud
To determine the optimal design practices for the Telco Cloud-Native Platform, this document adopts a top-down approach that profiles the workload across two-dimensional categories:
- Category A—Network function applications packaging, with the following options:
- Monolithic vAPP network function
- Cloud-ready VNF virtual network function
- Cloud-native CNF container network function
- Category B—Infrastructure Service-Level-Object (SLO), which we classify into three t-shirt sizes:
- General Purpose, that is, Compute Intensive
- Network Intensive
- Network and Accelerator Technology Intensive
The advantage of Workload profiling is essential as it helps telcos optimize resource allocation, capacity planning, cloud governance, and performance management. Below are some of the profiling parameters defining a telco workload:
- Resource requirements—Workload profiling assesses the CPU, memory, storage, and network bandwidth requirements of VNFs and CNFs. It helps find the resource footprint of each function under different operating conditions, such as peak loads or specific traffic patterns. Profiling can show the resource-intensive components of the network functions and guide resource allocation decisions.
- Performance metrics—Workload profiling aims to measure the performance metrics of VNFs and CNFs, such as throughput, latency, packet loss, and response times. By benchmarking the performance under various workloads, telcos can find bottlenecks, optimize configurations, and ensure that the network functions meet the required performance targets.
- Traffic patterns—Workload profiling analyzes the traffic patterns and characteristics that VNFs and CNFs handle. It helps find the variations in network traffic volume, types, and patterns. Profiling can highlight peak usage periods, burst traffic, or specific traffic profiles, allowing telcos to provision the right resources and scale the network functions accordingly.
- Scalability and elasticity—Workload profiling helps assess the scalability and elasticity requirements of VNFs and CNFs. It helps determine the behavior of the network functions when the workload increases or decreases. Profiling can find the scaling thresholds, the impact on performance, and the resource requirements to keep wanted service levels during workload fluctuations.
- Dependency analysis—Workload profiling helps in understanding the dependencies and interactions between VNFs and CNFs. It helps find dependencies on shared resources, potential contention issues, or performance impacts due to interactions between network functions. Profiling can guide the design of optimized network function chains and aid in resolving potential performance bottlenecks.
- Load balancing—Workload profiling helps decide the optimal load-balancing strategies for distributing network functions across different servers or clusters. Profiling helps in finding the load distribution patterns, resource utilization levels, and load-balancing algorithms that can maximize resource efficiency and ensure workload distribution across the infrastructure.
- Capacity planning—Workload profiling plays a vital role in capacity planning by providing insights into resource utilization trends and predicting future resource requirements. Profiling helps telcos estimate the necessary resources, plan for future growth, and optimize infrastructure investments to meet the anticipated workload demands.
- Performance optimization—Workload profiling enables telcos to find performance bottlenecks, optimize configurations, and fine-tune the network functions for improved efficiency. Profiling can guide parameter tuning, cache optimization, and other performance-enhancing techniques to achieve better resource utilization and higher performance levels.
The graph below is an illustrative example of workload demand analysis, highlighting the multitude of functions expected to co-exist within the Telecom landscape. Independent and network function equipment vendors deliver and package these functions in diverse software formats. However, we have successfully aligned this diversity with standard infrastructure guidelines, resulting in a streamlined approach that categorizes the underlying infrastructure into a maximum of three “t-shirt sizes.”
Figure 8. Network function workload profile analysis example