Guidelines for Managing Virtual Machines

Here are some guidelines for managing VMs:

• Use labels and descriptions. This will help you identify the purpose of an instance and also help when filtering lists of instances.
• Use managed instance groups to enable autoscaling and load balancing. These are key to deploying scalable and highly available services.
• Use GPUs for numeric-intensive processing, such as machine learning and high-performance computing. For some applications, GPUs can give greater performance benefit than adding another CPU.
• Use snapshots to save the state of a disk or to make copies. These can be saved in Cloud Storage and act as backups.
• Use preemptible instances for workloads that can tolerate disruption. This will reduce the cost of the instance by up to 80 percent.