Content for  TS 28.104  Word version:  18.3.0

This MDA capability is for analysis of resource utilization of 3GPP NFs.

The 3GPP system is a resource limited system, no matter whether the NF is working on virtualized resources or physical resources. Resource shortage would affect the QoS and potentially impact users' quality of experience (QoE), e.g., by lowering the users' data throughput, prolonging the users' data delay, raising the rejections for the establishment of new connections (e.g., RRC connection), sessions (e.g., PDU session) and resources (e.g., QoS flows, DRBs, etc.) and increasing the drops of the existing connections, sessions, and resources. This may also consequently lead to risking or failing SLAs. On the other hand, resource excess would cause wastage that leads to additional CapEx and OpEx. Therefore, it is imperative to ensure optimum and efficient resource utilization for the NFs. The resource utilization of an NF is heavily dependent on load or traffic patterns, which could vary in different coverage areas (e.g., business area, entertainment area, and residential area) and in different time periods (weekdays and time of the day). It is desirable that the spare resource of the low-usage areas can be allocated to the busy areas.It is expected that MDA can perform an analysis of the resource utilization for physical resources or virtualized resources for the 3GPP NFs (in a specific domain or cross domains) to indicate the resource usage patterns in the past and predict the resource usage trend for some time periods in the future. The physical resources to be analyzed may include hardware resources (e.g., CPU), DL and UL PRBs (for gNB), etc., while the virtualized resources to be analyzed may include virtual CPU, virtual memory, virtual disk, etc.It is also very useful that MDA correlates the resource analytics across 3GPP NFs and provides recommendations that can be utlized to efficiently orchestrate the resources among NFs between the low usage and high usage areas for some time periods. The recommended actions could be for example to optimise, i.e., increase or decrease the capacity of gNB to enhance allocation of the physical resources or to schedule the "scale in" and "scale out" of VNFs via ETSI MANO system to optimize the allocation of the virtualized resources.

This MDA capability is for analysis of control plane congestion.

As described in TS 23.501, a 5GC NF can become overloaded when it is operating over its nominal capacity resulting in diminished performance (including impacts to handling of incoming and outgoing traffic). Some mechanism, such as control plane congestion control as described in TS 23.501 is designed for the purpose of avoiding and handling of 5GC NF overload. For example, as described in clause 5.19.7 of TS 23.501, when the AMF is under overload conditions, it may reject the received request from the UE depending on various aspects. And the UE will send a new request after some time. It is possible that the new request will be rejected again because of the load of the AMF. In virtualized environment, the signaling request may be rejected due to inadequacy of available resources at the target 5GC NF e.g. AMF or SMF. If such situation can not be resolved, it will probably cause signalling storm for the whole network and affect the services (e.g. calls and data connections) provided by the network. It is desirable to use MDA to assist control plane congestion analysis in order to detect, prevent or resolve identified congestion issue happened at the control plane. MDAS producer may utilize the collected PM, FM, network topology data, virtual resource information provided from ETSI NFV MANO and etc.) for control plane congestion analysis and provides analytics report containing identified or predicted congestion issue for the target 5GC NF (e.g. AMF, SMF). The analytics report also provides recommended actions to optimize the target 5GC NF for avoiding or resolving congestion issue. Based on the recommendation in the report, 3GPP management system can adjust (e.g., scale-up the virtual resource) the resources to better facilitate processing of the received control plane messages.