Affinity Group

Description

An Affinity Group is a concept defined in 3GPP specifications 26.501 and 26.510, which cover 5G media streaming architecture and 5G media streaming protocols respectively. It refers to a set of service locations that are intended to be deployed with each other and not with members of other affinity groups. In essence, it expresses a deployment constraint: certain service instances or application endpoints belong together and should be co-located or co-selected, while instances from different affinity groups should remain separate.

Architecturally, the affinity group concept appears in the context of edge application server discovery and service hosting for media services. When a UE or a media client needs to access a service, the network must select an appropriate service location — typically an edge application server or a media function instance. The affinity group provides a way to express that certain locations form a cohesive set. For example, a media streaming service might deploy a set of edge servers that share caches, state, or session context; these servers form an affinity group. A client should be directed to a member of that group, and the group members should be deployed together, not mixed with servers from a different service or tenant.

The mechanism works through provisioning and discovery. The service provider or network operator configures affinity group information, associating service locations with group identifiers. During discovery, the network uses this information to filter candidate service locations, ensuring that the selected location belongs to the correct affinity group. This prevents misrouting a client to a server that is functionally incompatible or that belongs to a different deployment domain. It also supports consistency: if a client is moved or reconnected, the network can attempt to keep it within the same affinity group to preserve session continuity and cached state.

Key components include the affinity group identifier, the mapping of service locations to affinity groups, and the discovery and selection logic that enforces group membership. The concept interacts with edge computing enablers such as Edge Application Server discovery, with media streaming functions such as the Media Session Handler and Media Stream Handler, and with deployment and orchestration systems that place service instances.

Affinity groups matter because they give the network a standardized way to understand deployment topology constraints that are not visible from IP addresses or geographic location alone. Without this concept, service selection might choose a location that is topologically close but logically wrong, leading to broken sessions, cache misses, or cross-tenant interference. By encoding affinity, 3GPP enables correct and efficient service placement for media and edge applications.

Purpose & Motivation

The Affinity Group concept exists to solve a service placement problem in 5G media streaming and edge computing: not all service locations are interchangeable. Some locations are designed to work together as a set, sharing state, caches, or session context, and should not be mixed with locations from other deployments. Before this concept, service discovery and selection relied primarily on geographic or network proximity, which could direct a client to a server that was close in latency but incompatible in function or tenancy.

The motivation came from the growth of edge-hosted media services, where operators and content providers deploy multiple instances of media functions across edge sites. These instances often form logical clusters — for example, a cluster serving a particular content package or a particular enterprise. The affinity group provides a standardized label for such clusters, allowing the network to enforce co-deployment and co-selection constraints.

By defining affinity groups in 26.501 and 26.510, 3GPP addresses the limitations of purely location-based selection. It enables correct routing, preserves session continuity when clients move between edge sites, and supports multi-tenant deployments where isolation between groups is required. This makes edge media services more reliable and easier to operate at scale.