Over the past two years, the major native cloud players have launched cloud management product teams focused on improving the visibility, billing, and optimization capabilities available natively. Google has also joined this effort.
The company provides customers with cost and billing support via onboarding checklists, how-to guides, billing support specialists, and providing billing account, catalog, and budget APIs. Although Google bills customers on a monthly basis, it's also providing options for per-second, on-demand pricing in addition to various committed or sustained-use discounts:
- Sustained-use discounts. For standard virtual machines (VMs) being used more than 25% monthly, Compute Engine will automatically discount your bill. Better yet, if an instance is run for the entire month, the discount increases to 30%, an added benefit that provides a discount outside the traditional path of prepaying usage time.
- Committed-use discounts. Similar to Amazon Web Services' Reserved Instances, the user can receive a discount (up to 57%) in exchange for committing to using an instance for a certain period of time. When used optimally, users can save up to 57%. No upfront payment is necessary, and there is flexibility to change during the commitment period. The instance purchase is tied to a region, however, but can be changed to any instance configuration with few limitations.
- Preemptible VMs. Like Amazon's Spot Instances, Google Cloud Platform (GCP) offers a fixed 80% discount for virtual machine usage, though only a 30-second warning is provided before a preemptible instance changes price.
As is the case with all major cloud platforms, navigating billing and cost optimization gets highly complex. New users can kick off their savings with Google's free cost management tools (GCP Pricing Calculator, resource hierarchy, Cloud Identity and Access Management, or quotas). Keep in mind that with its per-second billing, anyone interacting with Google resources should create a script that starts, stops, and destroys a resource for the duration of the application usage. Not only will this prevent unnecessary costs but it could also prevent operational and security issues.
This post was written by Senior Analyst Tracy Woo, and it originally appeared here.