An organization runs fault-tolerant batch data processing and CI/CD build tasks on Google Cloud Compute Engine. The DevOps team needs to optimize compute expenses using Spot VMs while satisfying the following operational criteria: The fleet must use the Spot VM provisioning model to achieve up to 91% cost savings. The instances must gracefully handle preemption events to persist execution state and progress. The infrastructure must automatically recreate and resume instances when Spot capacity becomes available without manual intervention. The instance configuration must adhere to Google Cloud platform restrictions for Spot VMs. Which configuration strategy should the DevOps team implement?

Create an instance template with the scheduling provisioning model set to SPOT, omit automatic restart and live migration policies, configure a shutdown script to checkpoint job state upon preemption, and deploy the template within an autoscaled stateless Managed Instance Group (MIG).: What the Spot VM and Managed Instance Group Solution Is Spot Virtual Machines (VMs) provide excess Compute Engine capacity at discounts of up to 60–91% compared to standard on-demand pricing. Managed Instance Groups (MIGs) manage pools of homogeneous VM instances created from a shared instance template, providing automated lifecycle management, autohealing, and scaling capabilities. How it meets the Requirements Maximum Cost Optimization: Setting provisioningModel: SPOT (or --provisioning-model=SPOT) in the instance template configures all provisioned worker nodes as Spot VMs, maximizing cost efficiency for batch computing. Preemption Handling and Checkpointing: Spot VMs receive a preemption notice prior to termination. By embedding a shutdown script in the instance template metadata, the VM detects the shutdown signal and executes clean state saves or data checkpointing to persistent storage before stopping. Automated Lifecycle & Resumption: When Compute Engine preempts a Spot VM in a MIG, the MIG manager automatically monitors capacity and recreates or restarts the instance once Spot capacity becomes available in the target zone or region. Compliance with Host Maintenance Constraints: Spot VMs do not support host maintenance live migration or automatic infrastructure restarts; omitting these settings ensures the template strictly complies with Compute Engine availability policy rules. Key Advantages Resilient Batch Processing: Checkpointing via shutdown scripts prevents loss of compute progress. Declarative Fleet Management: Allows DevOps engineers to manage Spot infrastructure through Infrastructure as Code (Terraform) templates. Elastic Capacity Recovery: Managed Instance Groups dynamically reconcile desired capacity without operator overhead. FinOps Alignment: Drastically cuts operational compute expenditure on fault-tolerant asynchronous workloads. Why This Is Optimal Combining Spot VM provisioning with stateless MIG orchestration and shutdown script checkpointing delivers the ideal balance between high fault tolerance, automated recovery, platform compliance, and optimal cost reduction.