Unstructured Data Storage Function

Description

The Unstructured Data Storage Function (UDSF) is a component within the 5G Core (5GC) Service-Based Architecture, introduced in 3GPP Release 15. Its primary role is to provide persistent storage for unstructured data on behalf of other Network Functions (NFs). Unlike the UDR, which stores structured data profiles (subscription, policy), the UDSF is designed for data that does not conform to a predefined, standardized schema. This typically includes session context data, temporary state information, application-specific data blobs, or any other data an NF needs to preserve for reliability, scalability, or service continuity purposes.

Architecturally, the UDSF exposes a service-based interface, Nudsf, using HTTP/2 and JSON. The interface provides basic create, read, update, delete, and query (CRUDQ) operations. An NF (like a Session Management Function, SMF, or an Application Function, AF) can store a data object by providing a unique key (e.g., a combination of SUPI, DNN, S-NSSAI) and the data payload in a flexible format (often JSON or binary). The UDSF does not interpret the content; it simply stores and retrieves it. This makes the UDSF a generic storage service. A key use case is enabling stateless design for other NFs. For example, an SMF can store the Protocol Data Unit (PDU) session context in the UDSF. If the SMF instance fails, a new instance can retrieve the context from the UDSF and seamlessly continue session management, ensuring service continuity.

The UDSF operates independently of the data semantics. It is up to the producer and consumer NFs to agree on the data format. The UDSF may support features like data replication for high availability, access control based on NF identity, and potentially time-to-live (TTL) for temporary data. Its deployment is flexible; it can be a dedicated network function or its functionality can be integrated into a cloud-native data storage platform. By providing this service, the UDSF allows NFs to be designed as stateless or minimally stateful processes, which aligns with cloud-native principles. This simplifies NF implementation, enhances horizontal scalability (new NF instances can access shared state), and improves fault tolerance, which are critical for 5G's demanding service requirements.

Purpose & Motivation

The UDSF was created to address a specific challenge in the transition to a cloud-native, service-based 5G Core: efficient and reliable state management. Traditional telecom network functions were often stateful monoliths, storing session and context data locally. This made them difficult to scale horizontally and vulnerable to failures—if an instance failed, its sessions were lost. The 5GC design principle advocates for stateless NFs where possible, but some state (like PDU session context) is essential and must be persisted.

The UDSF provides a standardized solution for this persistence without mandating how NFs should structure their internal data. Previous approaches might have required each NF to implement its own proprietary database or rely on non-standardized storage, leading to complexity and vendor lock-in. The UDSF offers a simple, network-wide storage service. This solves the problem of state management for stateless NFs, enabling features like seamless service restoration after failures and efficient load balancing across NF instances. It is a key enabler for network slicing, as different slices can use the same UDSF infrastructure while maintaining isolated data storage for their respective NFs. The UDSF, therefore, is a foundational element for achieving the resilience, elasticity, and operational simplicity required by 5G networks.

Key Features

• Generic storage for unstructured data (e.g., session context, state blobs)
• Service-Based Interface (Nudsf) with CRUDQ operations using HTTP/2/JSON
• Enables stateless or minimally stateful design of other Network Functions
• Does not interpret data content; stores and retrieves based on key
• Supports service continuity and failure recovery for stateful processes
• Facilitates cloud-native principles like horizontal scaling and resilience

Evolution Across Releases

Defining Specifications