Today, a common initiative across the industry is increase agility and delivery while simultaneously decreasing infrastructure spend. As a result, we are seeing a massive amount of undeniable momentum building up behind the journey to the public cloud as providers like AWS enable organizations to unlock a more elastic and development-ready environment with the promise of decreased infrastructure cost. Although the elastic nature of the cloud is inarguable, the promise of decreasing IT investments is only possible if an organization can properly leverage and implement various methods of cost control. One of the ways to achieve decreased spend is to dynamically and autonomously scale the environment, both up and down, depending on utilization and workload demand. Another method which we will be discussing today, is to accurately invest in pre-paid capacity and the discounted rates these upfront investments will provide for your cloud bill.
What are Reserved Instances and how do they work?
Pre-paid capacity also known as reserved instances, within itself, is a wonderful pricing option that is offered by AWS. In short, AWS users have the a few options to submit upfront payments to “reserve” any number of EC2 templates for any specified period of time. To clarify, the “reservation” is not necessarily a physical reservation in the sense that you are blocking off certain instances/template types, but rather a reservation for discounted hourly rates that can be later applied to any live EC2 instance in your environment.
For example, if I have a workload that aligns with the specs of a m4.large template, and I know that this workload is relatively static in nature and rarely needs to be scaled, it would be in my best interest to pre-pay for this workload because of the added discounts that comes with the upfront payment.
To provide some numerical validation to this scenario; The On-Demand cost for an m4.large is $0.10 per hour. If you multiply that by the number of hours in a year (61,320), the yearly cost of this one instance would be $6,132. If I were to “reserve” this instance using the “All Upfront” option of AWS Standard 1 Year Term, I would need to pay an upfront cost of $507 but my hourly rate for this instance would drop to $0.058. If you then multiply the discounted hourly rate of $0.0508 by the number of hours in a year and then add on the $507 in upfront cost, the total yearly cost for this workload would only be $4,058.56 which is a 33.8% reduction in spend.
Figuring out what kind of Reserved Instances one should purchase can be quite complex and if the incorrect investments are made, an organization will end up spending significantly more than needed due to pre-paid Reserved Instances going to waste and failing to be applied.
What does it take to pick the right Reserved Instances?
In order to pick the best workloads for these upfront investments, one needs to first understand how every single workload fluctuates at various points throughout the year in terms of its resource demand. Secondly, most organizations have numerous compliance or business continuity policies in place that must be adhered to.
For example, some organizations have an initiative to be geographically diverse and resilient, or on the opposite end of the spectrum, some corporations are legally obliged to enforce stringent data locality regulations. Now, since the options for Reserved Instances in AWS are specific to each region, in order to generate these decisions, one must be able to analyze all the utilization data in relation to these compliance policies which turns this exercise from a challenging weekend hike into an ascension up Mt. Everest.
Is there an easier way?
With the release of Turbonomic 6.0, this painstaking task that often requires months of work, and in some cases outsourced consultants, can all be completed in a matter of minutes. The Turbonomic platform, out of the box, is built to fully understand all aspects of your environment including (but not limited to) resource utilization, volatility in workload demand, VM activity and uptime, as well as any compliance or business-related policies that induce constraints on where workloads must live. As a result of this holistic approach, the planning engine is able to accurately determine which workloads are best suited for Reserved Instance while adhering to all the policy constraints that may exist within your organization.
Within a few clicks in our plan tab you’ll be able to scope out any cloud migration scenario and will receive a full analysis on cost projections and comparisons across three different migration strategies (lift and shift, resize then shift, resize and leverage pre-paid capacity then shift). The best part is that the platform doesn’t stop once it gives you these cost values in the form of intuitive tables and graphs, the analysis will also provide you with workload by workload mapping in the sense that for each on-prem workload you plan to migrate, we will specifically list the best suited template if you were to lift and shift, the best suited template and its if you were resize then shift, as well as a checkbox to let you know if this workload is deemed to have stable enough utilization that a Reserved Instance can be leveraged for additional cost savings.
Long story short, with Turbonomic 6.0 you no longer have to break out your ice picks and saddle up for a two month ascension up Mt. Everest just to be in the cloud, you can just take an advanced autonomous helicopter right to the top. Also, users are now not only capable of leveraging autonomous scaling for performance assurance and cost control, but can also accurately determine and strategize which workloads should leverage reserved instances which in the one scenario outlined earlier, can result in cost reductions of 33.8% and sometimes even more. Enable your business to get the most out of every dollar invested and who knows what benefits your balance sheet will see by the end of the year!