Maximizing Cost Efficiency of Cloud Infrastructure

Contrary to a commonly held misconception, optimizing the value realized from public clouds, such as Google Cloud Platform (GCP), Amazon Web Services (AWS), and Azure, doesn’t necessarily mean minimizing spend on their services. In fact, growing cloud bills can be positive insofar as the consumer maximizes the value received per dollar spent, and the value realized exceeds that of alternative technologies. The goal should, therefore, not necessarily be to spend less on the cloud but rather to maximize the efficiency of cloud consumption. This is a critical distinction, especially for organizations that are in a growth phase and may be wary of any cost reduction initiatives that threaten to constrain operational flexibility.

Both growth- and margin-centric cloud consumers can maximize efficiency without inhibiting the attainment of their business goals. Here are some tips on how to do it.

  1. Create a migration business case for each workload migration or greenfield standup. Successful cloud deployments begin with careful planning, including detailed workload scoping, expense forecasting, and migration tracing. During this process, organizations should regularly compare actual expenses with projections to maintain discipline and to understand the causes of any variances encountered. They  should also update projections as necessary based on real-time learning from each migration phase. Therefore, until the organization migrates the last workload, the financial plan for a migration should not be “final.”
  2. Continuously evaluate workload placement. Getting the most value from the cloud means migrating (or starting greenfield) workloads that benefit from cloud-native technologies and/or those workloads expected to have highly variable consumption volumes. Organizations can reserve the data center for highly stable workloads that are not easily transformed into cloud-native or serverless architectures, such as generic workloads that are heavy in object storage, for example. If workloads are run in hybrid environments, organizations should constantly evaluate them for optimal placement. Additionally, as the nature of data center workloads changes, organizations should evaluate them for cloud migrations when and if they can be transformed or when usage patterns are expected to become highly variable.
  3. Build your organization’s FinOps capabilities. FinOps is the industry term for the collective set of capabilities supporting the management of spend on public clouds. Maximizing your team’s FinOps capability is every bit as important as maximizing its technical capabilities vis a vis architecting, securing, and maintaining the infrastructure itself. The primary pillars of FinOps center on maximizing the visibility and reportability of cost data, operational efficiency, and pricing efficiency.

Developing strong capabilities around the visibility and reportability of public cloud spend is by far the most important foundational pillar of FinOps. Metadata on billing line items such as tags, labels, or account naming conventions should be comprehensive enough that the organization can segment its cloud spend by business dimensions, better helping them understand their cloud economics. Examples of such segmentation might include segmentation by department, geographic region, customer segment, or by products and services. Although these dimensions will vary from organization to organization, there is one segmentation that nearly every organization must be able to implement: spend by technical owner. To meet this requirement, an organization should be able to trace every dollar of cloud spend to the person or people who incurred the cost.

Only with cloud spend comprehensively segmented can the other pillars of sound FinOps practices be applied. For example, with the ability to group cloud resources, technical owners can be consulted about the feasibility of downsizing or eliminating specific resources in their deployments. Similarly, cloud vendors offer a variety of commitment-based discount programs such as AWS Savings Plans and GCP Committed Use Discounts, the management of which falls under the pricing efficiency FinOps pillar. Optimizing participation in these programs requires accurate forecasts of future consumption by vendor SKU or product category. Compiling those forecasts accurately depends on an ability to track spend directly to technical owners so they can be consulted about their plans.

With careful planning, workload placement, and cost management, consumers of public cloud can maximize the value they realize from these services even when their bills are increasing month-over-month. So long as they are maximizing the value they get for those dollars, and the value received exceeds that offered by alternative technologies, rising cloud bills can be a very good thing.

By Rich Hoyer, Director of Customer FinOps at SADA