Skip to main content
Cloud & AI Hub
Browse
Glossary AI Directory Playgrounds Models Prompts Explainers Strategy Matrix Benchmark Decoder

Compute

Virtual machines, containers, and serverless execution.

21 terms

Compute is the layer where you choose how much control you want over the machine your code runs on, in exchange for how much operational work you take on. A VM gives you a full OS and everything that comes with patching one. A container gives you process isolation and a portable image without a full OS per workload. Serverless gives up the machine entirely — you ship a function and the platform handles provisioning, at the cost of cold starts and execution-time limits.

Most real architectures mix all three rather than picking one: a Kubernetes cluster of VMs running containers, with a few genuinely bursty or infrequent workloads pulled out into Lambda or Cloud Functions. The entries here cover the mechanics of each layer and — for containers specifically — the orchestration problem (Kubernetes) that shows up once you have more than a handful of them.