Computing metrics as a service (CMAS) for facilitating traffic steering in CATS framework

In the context of CATS applications, resource modeling and dynamic scheduling face core challenges: heterogeneous computing resources (e.g., CPUs, GPUs, FPGAs) with differentiated characteristics are difficult to unify through traditional coarse-grained metrics (e.g., virtual machine/container counts). Moreover, dynamically changing resource states (e.g., resource occupancy, service instance load cycles) complicate routing table maintenance in network nodes, creating bottlenecks for resources scheduling. This document provides a service-oriented computing capability modeling framework, abstracting heterogeneous resources into standardized service units for efficient resource modelling and traffic steering.