Intelligent Rightsizing for Google VMs and Persistent Disk to Reduce Waste

For the Google Cloud Platform (GCP), like all public clouds, you pay for the resources you provision regardless of how well utilized they are—or whether they’re used at all. If you don’t monitor your cloud usage, it makes it easy for waste to accumulate and costs to increase unnecessarily.

When looking for opportunities to optimize your cloud usage, it makes sense to begin with your biggest spend categories. For each of the major public clouds, the largest category is the virtual machines (VMs) and persistent disks (sometimes referred to as block storage) that back them. In fact, across our customer base this usage comprises 44% of GCP cloud bills.

Cloudability’s new rightsizing recommendations for Google Compute virtual machines (VMs) and persistent disk deliver powerful insights into where waste is accumulating and help you make intelligent trade-offs between cost and performance.

Maximize your rightsizing options by sourcing the right utilization metrics

Identifying the optimization options for any VM begins with understanding the historical resource demands for the workloads it hosts. At Apptio, we make it straightforward to source this utilization data either from the native monitoring tool or via a third-party agent that runs on each machine and pre-aggregates this information.

In the case of GCP, Cloudability connects to the native monitoring solution, Operations Suite, which makes available by default CPU, network, and disk utilization for all VMs. If you have Google’s Ops Agent or legacy Cloud Monitoring Agent installed on a VM, we’ll also source memory utilization data for that machine from Operations Suite. While memory data is not mandatory in building rightsizing recommendations, it does maximize the available optimization options and will help engineers take actions with confidence.

Consider opportunities to optimize with a custom machine

As with our rightsizing recommendations for AWS and Azure, we recognize that there is no such thing as the one perfect machine type or size to match a specific workload. The leading-edge rightsizing capabilities offered by Cloudability therefore aim to give you a minimum of five savings-risk ranked options per machine and interactive tools to understand the associated performance risk.

A unique aspect of GCP, and an exciting one from an optimization perspective, is the ability to craft a custom machine with a specific number of CPU cores and amount of memory. From a rightsizing standpoint, this gives us an opportunity to get really precise about matching a VM’s resources to the underlying workload. This is exactly what Cloudability does.

Typically, among the first five recommendations for a VM will be high-savings options for a zero-risk standard machine type and a zero-risk custom machine type. These will usually have different savings rates, but it’s important to provide you with the flexibility to choose what works for your organization. As always, explore all the options provided, overlaying each recommendation against hourly utilization data to build confidence before taking action.

Look for low-hanging fruit across persistent disk

The situation for optimizing the disks that back each VM is fairly similar across clouds. There are two main usage metrics that need to be considered—throughput and IOPS—and matched to a type of storage.

Our tip: there is usually some low-hanging fruit in terms of meaningful cost savings. Look for Cloudability recommendations that surface disks that are “extreme provisioned” or “SSD provisioned” but have small IOPS peaks. Another quick win is decommissioning disks that Cloudability identifies as completely idle. One way the platform makes this easier for you is locating the disks that have become unattached from their VM.

Operationalize your savings

We are excited to roll out our full set of rightsizing capabilities to GCP and see the cost savings made by our customers. Cloudability is a true multi-cloud platform and there are a number of common features to help you operationalize these cost savings. One of these is the aggregation of recommendations across projects and being able to leverage the filtered views system so stakeholders can see exactly what is relevant to them. To automate workflows, you may also like to create a policy specific to GCP rightsizing that creates tickets in your business systems.

Rightsizing recommendations for GCP are available to all Cloudability customers. Add the necessary credentials to the platform and you’ll be eliminating waste in no time.