The use of AWS, Azure, and other cloud services helps organizations to achieve critical business priorities. However, the complexity of cloud pricing combined with the rapid shift to cloud makes it difficult for CFOs to accurately forecast cloud costs. FinOps teams can help to better plan, estimate, and govern cloud spend, ultimately improving the accuracy of forecasts. FinOps teams must collaborate with finance and application owners to improve cloud cost forecasting while also ensuring efficient and optimized use of cloud resources.
Here are five common challenges FinOp teams face that drive inaccurate forecasts and cost overruns.
Challenge 1: Inaccurate forecasting when migrating to cloud
When you migrate on-premises workloads to AWS, Azure, or other clouds, you probably have access to historical data on the usage level for the application. However, you need to accurately inventory the on-premises resources used and translate those to equivalent cloud resources. During migration planning, many organizations neglect to consider the impact of software licensing costs, which can lead to inaccurate estimates of the application cost in the cloud. Rearchitecting workloads during the migration (such as leveraging PaaS services or moving to containers) can exacerbate the difficulties. It can also be tempting to overestimate your resource needs to prevent any risk of performance degradation.
Best practices for FinOps forecasting during migration
- Leverage migration planning tools to help automate your analysis and planning. Make sure you can fully map resources associated with your workload or business service.
- Rightsize as you migrate by identifying underutilized and overutilized servers on-prem. Avoid the temptation to over-provision.
- Compare prices across cloud providers to find the most cost-effective options. Consider cloud enterprise agreements, commitments, and savings plans.
- Don’t forget to plan for the cost of software licenses for cloud workloads, which can be higher than infrastructure costs. Explore BYOL options for long-running workloads which can have significant savings.
Challenge 2: Difficulty predicting cloud costs for new applications
When planning for new cloud applications, application and DevOps teams must balance costs, performance, and scalability. Often application owners don’t fully understand cloud costs until late in the development process. This creates budget overages and makes it difficult to adjust your application architecture to make it more cost-efficient.
Best practices for FinOps forecasting for new applications
- Engage your application owners on a regular cadence to identify new applications that will be deployed to cloud.
- Gain an understanding of how increased usage will impact costs for a new application so that you understand how costs might change in different usage scenarios.
- Compare costs on different cloud providers. Many organizations deploy workloads on AWS and Azure. Understand how costs will vary by cloud provider based on your existing enterprise agreements and discount options.
- Start cost planning early. FinOps teams should ensure application teams estimate cloud costs from the earliest stages. Require business owners proposing new applications to provide cloud cost estimates for business cases to be approved.
- Continue to refine cloud cost forecasts as you build the application and refine architecture and technical choices based on cost estimates. Technical architecture decisions, such as the use of cloud PaaS services, can have a major impact on operational costs.
Challenge 3: Complexity of governing cloud discounts
As cloud spend continues to grow, organizations increasingly use a variety of discount options to reduce their cloud bill. Commitment discounts like AWS Reserved instances and Azure Savings plan are widely used and enterprise agreements often include negotiated discounts based on the overall level of spend with a cloud provider. The complexity of these overlapping discounts combined with the continual expiration of commitment discounts can result in unexpected costs that exceed forecasts.
Best practices for improving forecast accuracy for discounts
- Determine your approach to managing commitment discounts: centralized (by a FinOps teams) or decentralized (by application owners).
- Identify the level of commitment coverage you are targeting (i.e. 70% commitment coverage on long running workloads).
- FinOps teams should work with application owners to forecast application usage and plan for commitment coverage.
- Continuously monitor commitment coverage and utilization as key KPIs and alert when they fall outside of your desired range. Make any necessary adjustments by converting, selling, buying, or re-allocating commitments.
- Get ahead of commitment expiration. Alert the relevant stakeholders well in advance of the expiration date so there is adequate time to plan.
Challenge 4: Uncontrolled software license costs in the cloud
FinOps teams are starting to recognize the importance of managing software license costs, which can often represent 25% of the total cloud bill1. Cloud compute and database services can include operating systems and databases from commercial vendors (Microsoft, Oracle, Red Hat) as part of the service. However, it can be difficult to determine whether pay-as-you-go or bring your own license models are more cost-effective, and users often don’t know whether any existing licenses are available. The result is wasted costs and increased risk. Marketplace spend may also lack governance, resulting in unexpected spend. A clear strategy and governance of software licenses is critical to ensure accurate forecasts.
Best practices for forecasting software costs in the cloud
- Leverage tools that provide visibility into current cloud software costs, including marketplace spend, pay-as-you-go licenses (PAYG), and BYOL. Leverage trends to identify how those costs will change.
- Provide guidance and controls about when to purchase software through cloud marketplaces (e.g., to meet spend levels in enterprise agreements).
- Require cloud application owners to plan for and communicate any necessary software licenses in their planning.
- Collaborate with SAM or ITAM teams to identify available licenses for BYOL, so that you can forecast the costs of PAYG licenses as well as any additional licenses to be purchased for BYOL.
Challenge 5: Spending anomalies that blow your forecast
There are several situations where existing applications can have spending anomalies that blow your forecasts. Reducing cost anomalies can minimize variance between your cloud forecasts and your actual spend. First, demand for a particular cloud application can exceed your planning scenarios, leading to additional costs. Second, deployment mistakes can lead to expensive over provisioning of cloud resources. Lastly, technical changes can result in additional spend that is not forecasted.
Best practices for preventing anomalies from blowing your forecast
- Get early notification when spend is deviating from your budget or forecast. Leverage budget alerts that notify you when spend on an application or environment is expected to be higher than expected.
- Implement regular reporting delivered to application owners, which, combined with budget alerts, will let them fix any issues or adjust forecast before they go too far astray.
- Alert development teams when their spend is increasing or decreasing beyond a specified percentage. This allows them to correlate unexpected shifts in spend with recent changes.
- Put governance in place to control how and when high-cost services can be provisioned.
How Flexera helps
Flexera offers a comprehensive set of FinOps solutions that can help you migrate, track, govern, and forecast your cloud spend. Identified as a leader in both the Gartner® Magic Quadrant™ for Cloud Financial Management Tools and the Forrester Wave™ for Cloud Cost Management and Optimization, Flexera provides solutions that you need to use cloud services efficiently and cost-effectively.
See Flexera FinOps solutions in action
Sources:
1Calculated by Flexera based on average spend on software as a percentage of cloud bill across Flexera Cloud Cost Optimization customers.
