Google Cloud Platform (GCP)
Google Cloud Platform (GCP) is the suite of cloud computing services offered by Google. It runs on the same global infrastructure that powers Google’s consumer products, including Search, YouTube, and Gmail, and it is one of the three hyperscale cloud service providers, alongside Amazon Web Services and Microsoft Azure.
GCP is sold under the broader Google Cloud brand, which also bundles enterprise products such as Google Workspace. The platform itself launched in 2008 with the preview of App Engine, Google’s Platform as a Service (PaaS) offering, and reached general availability in 2011. Google has since layered compute, storage, data, and machine-learning services on top of that foundation.
The compute portfolio spans the cloud service stack. Compute Engine provides virtual machines, the IaaS layer. App Engine is the PaaS. Cloud Functions is Google’s Function-as-a-Service (FaaS) product. Cloud Run runs containerized workloads without a cluster. Google Kubernetes Engine (GKE) is a managed container orchestration service built on Kubernetes.
Google Cloud Run
Google Cloud Run is a fully managed compute platform that runs containerized applications on request. It is a managed hosting service. Google provisions, scales, and retires the infrastructure, and the operator supplies only the container image and a concurrency setting. There is no cluster to size, no node to patch, and no control plane to keep available.
Cloud Run is built on Knative, the open-source Kubernetes extension that Google originated, and it inherits the container orchestration model’s declarative, reconciling behaviour. Where Kubernetes asks the operator to run a cluster, Cloud Run hides the cluster entirely. A request arrives, the platform starts one or more instances of the container to serve it, and the platform auto-scales those instances with traffic. It can scale down to zero when there is no work. Billing follows the same per-use model as FaaS. The customer pays for the CPU, memory, and time each instance actually consumes, and an idle service costs nothing.
This makes Cloud Run a middle ground between FaaS and a full Kubernetes deployment. The workload is a container, so it can carry any language, library, or binary that fits the image, and it is not bound by the runtime and execution limits of a FaaS platform. The operator is spared the operational burden of running a cluster. The trade-off is the loss of control. The platform owns the scheduling, networking, and lifecycle decisions, and workloads that need persistent background processes, long-lived connections, or tight placement control fit a full orchestrator better.