Technology Portfolio Management (TPM) Best Practices

Overview

Cloud infrastructure over-provisioning is one of the most pervasive and most recoverable sources of waste in enterprise cloud technology spending. When engineers provision cloud resources, they typically provision for peak load rather than average load, for growth that may not materialize at the anticipated rate, or for a safety margin that seemed prudent at provisioning time but that creates chronic over-provisioning once the workload’s actual resource consumption pattern is established. Once provisioned, over-sized resources tend to remain at their initial size indefinitely because the cost of investigating and right-sizing them is attributed to the engineering team while the savings of right-sizing accrue to a cloud budget that the engineering team does not directly control. FinOps governance creates the accountability structure that aligns these incentives.

Best Practice

Establish a continuous cloud infrastructure right-sizing program that uses cloud provider cost optimization recommendations, workload performance data, and resource utilization metrics to identify over-provisioned cloud resources and execute right-sizing actions on a defined cadence. For compute resources: establish a utilization threshold below which a resource is flagged for right-sizing review, and a remediation process that evaluates whether the resource can be downsized, replaced with a more cost-efficient instance type, or terminated without operational impact. For database resources: evaluate whether production database sizes reflect actual data volumes and query performance requirements rather than initial provisioning assumptions that pre-dated the actual workload. For storage resources: identify storage with low access frequency that could be migrated to lower-cost storage tiers without operational impact, and storage associated with terminated applications or services that can be deleted entirely. For network resources: identify data transfer costs that could be reduced through architectural changes such as regional traffic optimization or caching.

Benefit(s)

Continuous cloud infrastructure right-sizing produces ongoing financial recovery from the over-provisioning that accumulates in cloud environments that are not actively managed for cost efficiency. The financial recovery is recurring rather than one-time: each right-sizing action eliminates a cost that would otherwise continue indefinitely. And right-sizing produces co-benefits beyond financial savings: right-sized workloads often perform better under their actual load profile than over-provisioned workloads that are sized for a peak load that rarely or never occurs, because instance types optimized for the actual workload characteristics outperform generically over-provisioned instances.