Compute Engine (GCE)
Scalable, high-performance Virtual Machines (IaaS) running on Google's global infrastructure.
Scalable, high-performance Virtual Machines (IaaS) running on Google's global infrastructure.
What is Compute Engine (GCE)?
Google Cloud Compute Engine delivers configurable Virtual Machines (VMs) running on Google's global fiber network. It offers predefined and custom machine types (up to 416 vCPUs and 11.8 TB RAM on M3 instances), local NVMe SSDs, extreme networking throughput up to 200 Gbps, and live migration technology that keeps your VMs running seamlessly during host hardware and security maintenance.
Decision Guide: When to Choose This Service
Use Compute Engine when you require full OS-level control (Linux/Windows), legacy software installations, custom kernel configurations, specialized GPU/TPU machine types, or lift-and-shift enterprise VM migrations.
Do NOT use Compute Engine for standard containerized web microservices (use Cloud Run) or container orchestration (use GKE) unless you explicitly need low-level hypervisor or OS control.
Core Architectural Pillars
General-purpose (E2, N2, N4, C3), Compute-optimized (C2, C2D), Memory-optimized (M1, M2, M3), and Accelerator-optimized (A2, A3 with NVIDIA H100/A100 GPUs).
Automatically maintain VM availability, scale up/down based on CPU/HTTP metrics, and execute zero-downtime rolling updates.
Google automatically migrates running VMs to another physical host during host hardware updates without rebooting or dropping connections.
Excess compute capacity offered at up to 60-91% discount with the caveat that Google can reclaim instances with 30s notice.
Tailor exact vCPU and memory ratios to your workload without paying for unwanted overhead.