COMPEL Glossary / cloud-native-architecture
Cloud-Native Architecture
Cloud-native architecture refers to systems designed specifically to leverage cloud computing capabilities such as elastic scaling, distributed processing, managed services, containerized deployment, and microservice decomposition.
What this means in practice
Rather than simply moving existing applications to the cloud (lift-and-shift), cloud-native systems are built from the ground up to be resilient, scalable, and efficiently managed in cloud environments. For AI workloads, cloud-native architecture enables organizations to scale training jobs to hundreds of GPUs on demand, serve models with automatic scaling, and pay only for resources actually consumed. In COMPEL, cloud-native architecture decisions are part of the Technology pillar assessment during Calibrate and the platform strategy design in Module 3.3, where build-versus-buy and vendor lock-in considerations are central to the AITGP's architectural recommendations.
Why it matters
Organizations that simply lift-and-shift existing applications to the cloud miss the scalability, resilience, and cost optimization benefits that cloud-native design enables. Cloud-native AI architecture allows elastic scaling for training jobs, auto-scaling model serving, and pay-per-use economics that make AI operations financially sustainable. The architectural decision between cloud-native and legacy approaches significantly impacts long-term AI capability and cost structure.
How COMPEL uses it
Cloud-native architecture decisions are part of the Technology pillar assessment during Calibrate and the platform strategy design during Model. Build-versus-buy and vendor lock-in considerations are central to the architectural recommendations. During Produce, cloud-native AI infrastructure is implemented using containerization, microservices, and managed services. The Evaluate stage measures whether the architecture delivers the elasticity and cost efficiency projected in the design.
Related Terms
Other glossary terms mentioned in this entry's definition and context.