Cloud spend continues to increase as the need to deliver a better customer experience (CX) — via AI, the metaverse, or as-a-service platforms — shows no signs of slowing down. Whether for a B2B or B2C customer or an internal IT or DevOps team, cloud-based applications are now a norm, and most organizations and teams are experimenting with generative AI capabilities. The problem is that digital-driven enterprises are enabling cloud spend to increase without subjecting it to scrutiny that can better rationalize the spending against business outcomes and ROI. Enter FinOps.
A distributed model empowers organizations
FinOps is a cloud financial management practice gaining favor as businesses realize they have a far less than ideal view of cloud costs and to whom to attribute those costs. Cloud operations often work in silos: a data science team, for example, is storing files in the cloud or retrieving data, while a development team may create a cloud-native application for a better CX. When the ultimate goal is considered, the costs tend to blur — if a data science team reveals a need for a revised or enhanced cloud application for CX, how should the costs be shared with the DevOps, data science, IT, and CX teams?
Businesses have used the conventional hub-and-spoke model to cost out cloud operations, relying on feeding usage data into a centralized FinOps service. It does not synchronize with modern organizations that use hundreds, if not thousands, of managed services. Centralizing cost optimization is no longer tenable. If Gen 2.0 focused on optimizing cloud "infrastructure" costs, Gen 3.0 needs to be all about optimizing "services" costs.
The recommended modern approach is to move to a distributed model. It’s gaining in popularity, since it enables businesses to deploy usage-based pricing as well as identify what services are generating costs and by how much. The distributed model helps fine-tune accountability by delving deeper into the data, streams, and insights to rationalize costs against service usage. It’s not just about instance types, networks, or storage anymore. For closed-loop systems to optimize costs based on service usage, this level of granular data is vital.
A best practice in the distributed approach is using machine learning (ML) modeling to identify resource usage patterns and reduce costs. To be most efficient, IT needs to train a model within a service to achieve the desired granularity. In one use case, ML can identify resource usage patterns and contrast native Azure Virtual Desktop deployment with a Parallels RAS Virtual Desktop and application delivery platform — the result is a 79% cost optimization.
ML can also detect unused or rarely used resources, which continue to be a significant source of cloud spend consumption.
Fostering a FinOps culture
Industry reports show that while FinOps acceptance is growing, there is still work to be done. The FinOps Foundation State of FinOps 2023 report notes 57.3% are using cloud utilization data, leaving room for more deployment. In addition to cloud usage data, businesses are adding other FinOps data sources to get a multifaceted view into cloud cost management. Other popular data sources are IT finance data (38.3%), general business data (28.9%), and performance and observability data (28.4%).
Like any major initiative, to succeed, FinOps has certain dynamics that must be present:
- Leadership buy-in: FinOps Foundation reports a shift to CTOs as the most likely reporting role, with CFOs gaining prominence and CIOs diminishing. A major challenge for CTOs is engaging teams in the FinOps culture: the top challenge reported (40%) is getting engineers to act on cost optimization recommendations.
- Creativity support: Development and engineering teams face constant time-to-market pressures and need the ability to make cloud services decisions and create applications/services without an unwieldy approvals process.
- Transparency: In a distributed FinOps model with the mission of sharing cloud costs, transparency is a must. CTOs, CFOs, directors, team leaders … all must embrace the value of visibility in cloud costs and communicate that message across the organization.
A promising beginning
FinOps is showing encouraging signs of more acceptance as technology and financial executives see the need to better rationalize their cloud computing against costs and ROI. There is a need for efficient growth in this macroeconomic condition that is surfacing FinOps to every service or product owner. Organizations will continue to move to a distributed FinOps model to best accommodate the diverse use of cloud-based services.
As new cloud services appear with terabytes of data (or more), distributed FinOps will provide the foundation by which businesses can make better strategic cloud budget decisions and strengthen ROI.