An organization is evaluating Google Cloud compute options for two distinct application workloads: Workload 1: A lightweight, event-driven HTTP service with spiky traffic patterns that requires zero server maintenance, automated scaling to zero, and minimal operational complexity. Workload 2: An enterprise multi-tenant processing engine that requires fine-grained pod-level network policy isolation, custom node OS configurations, and in-cluster role-based access control (RBAC). Which compute platform design meets these requirements while optimizing administrative overhead and infrastructure control?

Deploy Workload 1 to Cloud Run and deploy Workload 2 to Google Kubernetes Engine (GKE) Standard.: ### What the Recommended Architecture Is This architecture pairs **Cloud Run**—a fully managed serverless container execution environment—with **Google Kubernetes Engine (GKE) Standard**, a managed Kubernetes platform providing deep infrastructure control. ### How It Meets Workload Requirements - **Workload 1 (Serverless & Zero Maintenance):** **Cloud Run** abstracts all underlying server, node, and cluster operations. It automatically manages scaling up under load and scaling down to zero when idle, minimizing both operational overhead and idle compute costs for lightweight HTTP services. - **Workload 2 (Infrastructure Control & Custom Configuration):** **GKE Standard** gives administrators granular control over node pool configurations, allowing custom OS images (such as Ubuntu or Container-Optimized OS), specialized machine families, and custom VM sizing. Inside the cluster, administrators can enforce multi-tenant isolation using **Kubernetes Network Policies** to restrict pod-to-pod communication and **Kubernetes RBAC** (`Roles` and `RoleBindings`) to enforce fine-grained permissions across namespaces. ### Key Advantages and Capabilities - **Operational Efficiency:** Offloads infrastructure provisioning and management to Google Cloud for stateless, spiky services via **Cloud Run**. - **Granular Customization:** Allows custom node pool configuration, daemonsets, and underlying VM parameter tuning in **GKE Standard**. - **Advanced Multi-Tenancy:** Provides native Kubernetes primitives such as namespaces, resource quotas, and network policies to securely isolate tenant workloads. - **Optimized Total Cost of Ownership (TCO):** Avoids paying for idle compute on spiky workloads while reserving dedicated compute capacity where deep control is required. ### Why This Is Optimal Using Cloud Run for simple HTTP services removes cluster maintenance overhead (such as control plane version management and node pool sizing), while deploying GKE Standard for the complex pipeline provides the required host-level flexibility and internal Kubernetes access control mechanisms that serverless abstractions do not support.