In a recent post, we introduced Effective Avoidance Rate (EAR) to quantify financial outcomes resulting from workload optimization. We drew a parallel between EAR and Effective Savings Rate (ESR), and mentioned an interaction between them. Assuming cloud cost optimization involves a combination of rate and workload optimization, understanding the relationship between them allows us to measure the total cloud cost optimization outcome. Here, we examine each metric in more detail and propose a method for combining them to derive Effective Cost Optimization Rate (ECOR). This holistic ROI metric quantifies total cloud optimization.
Measuring Total Cloud Optimization
Challenge
While ESR and EAR are each useful in quantifying the impacts of rate and workload optimization, respectively, FinOps practitioners are ultimately concerned with the total impact of their combined cost optimization efforts. An interesting dynamic is that ESR and EAR can work in opposition to each other. Workload optimization reduces usage that can be discounted, while rate optimization involves commitments that are paid for regardless of whether the corresponding resources are workload-optimized.
How do you determine the right balance of rate and usage optimization that yields the greatest economic outcome? Higher ESR and lower EAR? Lower EAR and higher ESR? A moderate mix of each?
In rate optimization, the highest savings are realized by balancing discount coverage and utilization. This balance is achieved by optimizing ESR, a single metric that indicates the overall savings outcome. To find the ideal combination of rate and workload optimization, we need a single metric that indicates the overall economic outcome. This is ECOR.
Definition
ESR, EAR, and ECOR are all conceptually similar and derived from the same core formula applied in different contexts:
Applied to the rate optimization context, we get:
Applied to the workload optimization context, we get:
If we now apply it to the combined cost optimization context, we get:
Note: ECOR is not simply the sum of ESR and EAR; each metric represents the impact of the “other” optimization type, reflected in different denominators. We use a common, pre-optimization denominator to bring these metrics together.
Example
Let’s look at a simple example to make this real. Dynamic, at-scale environments are more complex, but this illustrates the fundamental concepts.
Over a 24-hour period:
- An instance costs $1.00/hr on-demand
- 7 Reserved Instances (RIs) exist, and cost $0.70/hr each
- For the first 12 hours, 10 instances run
- For the second 12 hours, 5 instances are stopped as part of a schedule
Effective Savings Rate
Actual Spend
- Includes the impact of both rate optimization (7 RIs) & workload optimization (stopping 5 instances)
- RI spend + Remaining on-demand spend
- 7 RIs x $0.70/hr x 24 hrs + 3 instances x $1.00/hr x 12 hrs
- $153.60
Pre-Rate Optimization Spend
- Costs that would have occurred without rate optimization
- Includes workload optimization
- 10 instances x $1.00/hr x 12 hrs + 5 instances x $1.00/hr x 12 hrs
- $180.00
Savings Generated
- Pre-Rate Optimization Spend – Actual Spend
- $180.00 – $153.60 = $26.40
Effective Avoidance Rate
Pre-Workload Optimization Spend
- Costs that would have occurred without workload optimization
- Includes rate optimization
- 7 RIs x $0.70/hr x 24 hrs + 3 instances x $1.00/hr x 24 hrs
- $189.60
Costs Avoided
- Pre-Workload Optimization Spend – Actual Spend
- $189.60 – $153.60 = $36.00
Effective Cost Optimization Rate
Pre-Workload & Pre-Rate Optimization Spend
- Costs that would have occurred without rate or workload optimization
- 10 instances x $1.00/hr x 24 hrs
- $240.00
This means the overall combined rate and workload optimization activities resulted in a 26% reduction of cost relative to doing nothing at all.
Conclusion
ESR and EAR are each valuable for measuring the independent effects of rate and workload optimization, respectively. ECOR allows us to more completely measure the total economic impact of cloud optimization, unifying rate and workload optimization. By maximizing ECOR, organizations can identify the balance of rate and workload optimization that creates the best overall economic outcome.
Next Steps
ESR has been adopted by the FinOps Foundation and FinOps community to quantify rate optimization efforts. EAR and ECOR are the companion metrics that enable us to measure workload, and overall, cloud optimization outcomes. Together, they can help practitioners understand, benchmark, and improve their optimization approaches.
We look forward to sharing more details and collaborating with the FinOps community to refine and standardize these KPIs.
Prosper on! 🖖
-Joe
