Cloud Cost Optimization: 10-day vs. 30-day Rightsizing
Rightsizing can lower your costs like magic, but you need to get it right. The problem is that it isn’t easy — or even possible — to rightsize without having accurate analytics to guide you.
Rightsizing is pretty much what it sounds like: the process of defining the optimal cloud infrastructure best suited for the current and near-term needs of workloads in a way that balances risk and cost to minimize waste.
Cloud resources are elastic, scalable and provisioned on-demand. It’s a combination that allows for precision rightsizing on a continuous basis, but only when done correctly. When determining which instances to rightsize, there are two key attributes at play: risk and savings. Under-provisioned instances can result in degraded application performance, poor user experience and, ultimately, negative revenue impact. Traditionally, organizations have adopted an approach of over-provisioning resources, choosing to favor technical over financial risk, ensuring performance and customer experience isn’t impacted.
With this in mind, cloud workloads are typically launched taking into account some assumptions. One assumption is the look-back window of time, or representative dataset, used to profile and predict your cloud infrastructure needs.
While we generally see that a 10-day look-back window is most beneficial, there are times where 30 days can be useful. Still, how do you select which option is best for you? Like most things, it really depends on the workload and use case.
When You Should Look at 30 Days
Across most use cases and scenarios, a 10-day look-back period captures the most recent performance trends and is more predictive of future resource use and growth. Look-back periods for 30 days can be helpful for the scenarios we’ve outlined below.
Monthly cyclical usage
The most common reason for leveraging a longer look-back window for rightsizing is when there is a monthly or longer cyclical nature of the business. Here we see usage and files grow larger as the month goes on, then returning to a lower level.
For these instances, infrastructure should be based on peak usage later in the month. Infrastructure sized appropriately when compared to the first 10 days of the month will appear underutilized and could skew recommendations toward less than optimal sizing. Examples of these type of workloads include (but are not limited to) billing, payroll and inventory.
Over 10-day windows between regular events
Some organizations have regular internal or external events that require increased CPU, bandwidth and memory with greater than 10-day intervals between sessions. Sizing according to a smaller off-peak window may have serious implications during events that require more resources, including having to purchase expensive peak on-demand capacity, degraded performance, and even loss of service. Examples of these types of workloads include live transaction processing, streaming events, gaming and science/research.
Organizational and operational requirements
Many organizations have operational procedure and process needs based on monthly cycles for audit, verification, maintenance or finance. While the recommendations between 10-days and 30-days for these situations don’t tend to differ significantly, the ability to easily toggle between a 10-day and 30-day interval can offer additional peace of mind that recommendations include adequate coverage since a previous audit/review cycle.
Track and optimize public cloud spend with Cloudability
Cloudability rightsizing enables users to easily switch between a 10-day and 30-day window for different use cases, departments, workloads or just peace of mind. Whichever you choose, our platform leverages over a trillion hours of usage data, industry-leading algorithms and machine learning to continually generate recommendations ranked across several risk profiles to optimize your infrastructure.
Prioritized recommendations, estimated savings, estimated optimized spend, clipping risk projections and resource utilization (CPU, network bandwidth, memory and disk) all adjust in real time to help you optimize your infrastructure.
For API users the change of an argument to an endpoint returns recommendations based on the desired look-back window.
Ready to rightsize your infrastructure? Download our Rightsizing Tech Brief to learn more about Cloudability’s Rightsizing Recommendation Engine for AWS and Azure or reach out to one of our experts directly.