In our last blog post we touched upon how Reserved Instances generally offer more flexibility, freedom and of course, more significant savings than On-Demand Instances do. Let’s now take a closer look at how that actually plays out and how you can take advantage of the opportunity Reserved Instances provide.
The Real End of Your Reserved Instance Term
When you sign up for a Reserved Instance, while you might have a 1 or 3 year term, your actual lock-in will most likely end far before that. So despite the perception that reservations increase your commitment, your break-even point is the true end of your lock-in and commitment. This justifies – from a financial perspective – your freedom to either continue using the Reservation or opt for a different package/vendor.
To illustrate this, in the graph above you can clearly see for the Linux m.1 large instance, that the orange line – the Light Utilization – crosses and drops below the blue On-Demand line – at around 3 months. This means, that the ROI point of the Light Utilization is 3 months, or ~8% utilization. As such, even for a short-term, 3-month project, the Reserved Instances are more cost-effective than On-Demand.
Additionally, even if you are only using Light Utilization instances for spikes, it’s still a better pricing option than On-Demand, even if the instance runs only 3 month out of 36, or 8% of the time!
Now let’s also take a look at a t1.micro instance in the Singapore region.
The behavior here is entirely different with the ROI point of the Light reservations at ~11 months (~32% utilization) and the Light and Medium reservation lines cross the on-demand line almost at the same time. So for this instance, the Light reservation is never worthwhile! But you could still save over 30% for the 3-year term with the Medium reservation vs. On-Demand – giving you the freedom to stick with the reservation or move to another pricing plan or vendor.
This Sounds Too Easy
So if Reserved Instances deliver such amazing benefits, IT and finance departments should be all over it. But as we pointed out in our previous post, the majority of companies do not use reservations.
To explain this, let’s look at the two examples above. They clearly show that while reservations are generally more cost-effective than on-demand, you cannot simply assume that all reservations are the same. Factors such as region, instance type and platform can significantly impact the cost and ROI points.
Additionally, while the break-even calculation is straight-forward for a single instance, things get very complex when dealing with multiple instances. You need to calculate the number of instances of each type running simultaneously and the percentage of utilization time to provide context for any forward-looking recommendation. And this is simply unfeasible without some sort of automation tool.
Simplifying Your Cloud for Maximized ROI
Cloudyn’s system automatically performs this comparative analysis based on your actual consumption trends, with highly accurate cost-saving recommendations. As you can see in the screenshot below, we show the number of recommended instances to reserve, the estimated on-demand cost, the estimated cost of reserved instances, all making it very easy to see the impact of any recommendation. (The cost figures also take into account the percentage of time the instances are expected to stay up, and not rely on 100% utilization.)
To discover how you can begin optimizing your cloud consumption – sign-up free today!