Rate Optimization

Driving cloud rate efficiency through a combination of negotiated discounts, commitment discounts (RIs, Savings Plans, Committed Use Discounts), and other pricing mechanisms to meet the organization’s operational and budgetary objectives.

Definition

Rate Optimization is the Capability that helps you to lower the rate that you pay for the resources you use in the cloud. Cloud cost is based on the amount of a resource you use multiplied by the rate that you pay for it. Organizations manage rates by negotiating discounts with cloud providers or other vendors of consumption-based services, by purchasing commercially-available, resource-based commitment discounts (Reserved Instances, Committed Use Discounts), by purchasing spend-based commitment discounts (Savings Plans), or by taking advantage of usage-based discounts, special program discounts, or interruptible resources such as Spot compute instances.

Each cloud service provider has different approaches to negotiated discounts and special programs which affect rates. Negotiated—or organization-specific—discounts are often private, covered by non-disclosure rules, and not shared publicly. Be sure to understand your contractual obligations with respect to any of your agreements and follow them. Typically these programs will be managed by the Procurement persona. The FinOps team will play a key role in providing data to inform Procurement about historical and planned cloud usage, and to highlight important cloud cost and business value drivers.

The most important rate management mechanisms are spend-based commitment discounts, primarily Savings Plans, and resource-based commitment discounts, primarily Reserved Instances (RIs) and Committed Use Discounts (CUDs).

Spend-based commitment discounts and resource-based commitment discounts are the most popular rate optimizations that cloud service providers offer. These discounts allow organizations to commit to using certain resources, resource categories, or an amount of spending over a 1 or 3 year period in exchange for a published, discounted rate. Organizations should evaluate the level of consistently-running resources coverable by a certain discount, and establish purchasing strategies to reach coverage or effective savings rates appropriate for the variability and financial posture of the organization’s cloud use. Discounts are higher the more restrictive the mechanism. A 3 year commitment of a particular size of compute resource in a particular geography will yield a higher discount than a 1 year spend commitment that would cover usage of many compute types in any geography.

A variety of models of RIs, CUDs, and Savings Plans are available to cover Compute resources, but cloud providers also offer these discounts for other types of services as well. They are often more restrictive, but can also offer higher discounts. FinOps teams should consider all rate discount options for the services they use.

An organization’s central FinOps team should be the primary driver coordinating purchases of spend-based and resource-based discounts. These discounts are complex to understand, have many parameters, and can be purchased in ways that allow their benefits to cover resources throughout the organization’s cloud estate. Committed discounts also can affect the benefits of doing other types of optimization such as Workload Optimization, Licensing and SaaS use, changing cloud architectures, or cloud sustainability recommendations. The central FinOps team is best positioned to understand all of these optimization options and their interactions and coordinate the best approach with the appropriate personas in any given case. The central FinOps team may not make all the purchases, particularly for RIs for non-compute resources that may only be used by one or a few applications, but they should be tracking all of these purchases on behalf of the organization.

Cloud Providers and many FinOps platforms and tools include recommendation engines that allow organizations to plan, manage, and track benefits from these types of discounts. There are also vendors who specialize in managing commitment discount portfolios, even as a managed service. The FinOps Landscape allows organizations to search for FinOps tools and service providers appropriate to this task.

Other rate optimization can include the use of interruptible virtual machines (mostly called Spot instances), which each of the clouds offers. Spot instances are essentially spare capacity offered at a discounted rate where the cloud provider may recall the instance if purchased by another user at a non-spot rate. Use of spot instances for compute has important implications to the architecture of the systems using them, so FinOps team should consult with Engineering teams to consider use of Spot, and assist with understanding where it will provide value in excess of the risk and architectural work required. Note: Spot can also be considered a workload optimization as it does involve changing the workload to support and utilize the offering.

Rate Optimization has a very strong interrelationship with Workload Optimization. Recommendations to purchase commitments for resources to reduce their rate may conflict with recommendations to change the type or size of resource running in a system. This causes two primary problems. First, if a FinOps team is looking at the potential benefit of taking actions in either category, they may double count potential cost avoidance. Second, if a FinOps team takes action on both of these recommendations they may end up having to pay for a year for resources that no longer exist in their original form. It’s important to note that commitment discounts are not resources. They are similar to coupons that can cover certain types of resources that are running. If no resource is running which matches the type of the commitment, the organization still pays for the resource.

In an ideal world, all workload optimization would be done prior to rate optimization actions, but in reality there are constraints on engineering teams that can make that difficult. The FinOps team will be responsible for doing enough of both to strike the balance of value creation for the organization. It is important to avoid waiting for workloads to be fully optimized prior to considering rate optimizations as this will almost always lead to paying more than a balanced approach.

Similar conflicts may exist between Cloud Sustainability and Rate Optimization where buying reservations for resources that are allowed to run in an unoptimized state for years may increase carbon impact unnecessarily. An optimal approach only looks to purchase rate discounts for the level of resources that is actually required at any given time.

Maturity Assessment

Functional Activities

Measures of Success & KPIs

• Ability to measure the overall effective savings rate of Cloud Rate Optimization efforts for both technology-based and monetary-based commitment discounts
• For resource-based commitment discounts, maintaining a utilization upper-waterline around 80% for steady-state usage
• For spend-based commitment discounts, purchasing commitments with at least 90% savings per dollar of commitment for an established threshold of peak variable usage
• Analysis and purchase decisions for commitments are made in the context of interruptible/batch/highly variable workloads
• Ability to identify unused commitment based discounts with daily resolution
• Ability to notify stakeholder teams about expiring commitments with sufficient time to plan a new purchase
• Purchasing of commitment based discounts are viewed as investments by stakeholder teams like Execs/Productment/Finance; the investment is the cost of the commitment over the entire period, and the return is the savings provided
• Hybrid purchasing strategy that Aligns commitment terms with infrastructure workload characteristics and lifecycle
• Purchasing commitments that deliver more than 10% return on investment
• Mitigate risk by purchasing commitments with a break-even within 9 months
• Analysis and management is done centrally using a holistic view of the organization’s cloud estate and not at each individual cloud sub-account level
• Analysis for making commitment purchases is supplemented with planned infrastructure and/or workload capacity changes
• Commitment purchases are spread over the year to allow for flexibility by always have some % of commitments expiring; this enables re-evaluation of commitment levels at regular intervals informed by forecasted future usage
• Analysis and purchase decisions for commitments are made in the context of any negotiated commercial discounts offered to enterprises by the cloud service provider in exchange for overall cloud spend
• Altogether, the implementation of these strategies drives an organization’s Effective Savings Rate (ESR). It is important to note that under utilization of a commitment based discount would also negatively impact ESR as would significant usage not covered by discounts.

Inputs & Outputs

Inputs

• Workload Optimization recommendations
• Rate Optimization recommendations
• Architecting for Cloud recommendations
• Cloud Sustainability recommendations

Outputs

• Rate Optimization Strategy
• Rate optimization recommendations