FinOps is a new discipline focused on managing and optimizing cloud costs. It combines “Finance” and “Operations” with the goal of saving money and maximizing the business value of the cloud.

What FinOps Involves

FinOps requires a shift in mindset and close collaboration between finance teams, IT teams, and the business. Knowing what you’re going to pay is important. The ultimate goal is real-time visibility into cloud spending, so you can make adjustments along the way. FinOps focuses on identifying cost-saving opportunities, optimizing resource usage, and negotiating better rates with cloud providers. It also promotes transparency and financial accountability within teams.

Total Cost of Ownership (TCO)

Total Cost of Ownership (TCO) in the cloud isn’t just about the monthly bill—it includes all direct and indirect costs associated with running your applications and infrastructure in the cloud compared to on-premises. Optimizing TCO is an ongoing process that requires a strong focus on FinOps best practices.

Below are some best practices for optimizing Cloud TCO:

• Foster a FinOps Culture
  • Encourage collaboration between engineering, finance, and business teams.
  • Use a shared responsibility model where product teams and business units are accountable for their cloud spending.
  • Train teams on best practices and how their actions impact costs.
  • Appoint a dedicated FinOps team or leader.
• Gain Detailed Cost Insights
  • Allocate costs to specific teams, projects, or products.
  • Use tools like AWS Cost Explorer, Azure Cost Management, or Google Cloud Cost Management, and consider third-party FinOps platforms for more advanced analysis.
  • Ideally, maintain a single, consolidated view of all spending across a multi-cloud environment.
  • Aim for hourly (or even finer) granularity in your cost data to identify spikes.
• Optimize Resource Usage
  • Analyze resource usage and adjust dynamically.
  • Automatically shut down idle resources.
  • Auto-scaling can help match resources to demand, but it may also make costs less predictable.
  • Minimize data transfer between regions or availability zones, and use Content Delivery Networks (CDNs) for static content.
• Strategic Use of Pricing Models
  • Negotiate with your cloud provider for long-term commitments (1 or 3 years) for stable workloads to secure discounts.
  • Explore volume discounts and negotiate custom pricing if your cloud spend is significant.
• Implement Governance and Automation
  • Set up alerts to notify stakeholders when spending approaches or exceeds thresholds.
  • Use automation tools and scripts to rightsize resources, shut down inactive environments, apply tags, and enforce policies.
  • Monitor for unexpected spending spikes and investigate root causes as quickly as possible.

Egress

Egress costs are the charges cloud providers apply for transferring data out of their cloud environment. These costs can be significant. Examples of egress include data pulled from a website hosted in the cloud, transferring data to another cloud provider, communication with your on-premises data center, and even pushing data to another region or availability zone (these costs are usually lower).

AI

The use of AI is likely indispensable in a modern organization. Unfortunately, using AI presents major challenges. Those using generative AI (like ChatGPT) may already be familiar with tokens used per prompt. Organizations training their own AI or using machine learning may face higher costs for GPUs and for data usage or storage.
On the other hand, we are already seeing the emergence of ‘AI-powered FinOps’, where AI helps us fine-tune and optimize operations!

Conclusion

FinOps is an important discipline for organizations aiming to optimize their cloud costs. A key first step is fostering collaboration and accountability in this area. That also requires clear and up-to-date visibility into costs. By continuously monitoring and adjusting, organizations can then achieve meaningful savings.
Keep in mind that reducing costs isn’t everything. If you force teams to use fewer resources, it could lead to decreased quality for both internal and external customers—and that’s definitely not the goal!

Onno Poelmeyer, Consultant, CoolProfs