AWS Reserved Instances: Don’t Just Start Smart, Stay Smart
In our previous blog we talked about starting smart: how HyperCloud Analytics recommends the appropriate instance type for your application. This blog we cover the topic that most organizations are oblivious to: continuous cost optimization. In other words, how do you stay smart with your Reserved Instance (RI) choices.
How do you ensure you proactively manage costs before it becomes a problem to solve. and not let it come to the stage where you have to become Grampa Simpson below –
HyperCloud 6.0, comes with an extended “Start Smart, Stay Smart” cost optimization feature to leverage AWS Reserved Instances. This will help you understand when RIs can be used to reduce or bill and just as importantly, show you where unused RIs are adding to the wastage in your account.
Finding Hidden Savings
To get started with RIs, look at your EC2 estate, and find opportunities for saving via reservations. There are typically plenty of low-hanging fruits, long-running instances that have not been changed in a long time. But, visual inspection doesn’t work at scale.
HyperCloud 6.0 can help you pick from curated RI options based on deep insights from your historic cloud and instance usage patterns and real time price vs. performance.
For those of you new in the RI game, we recommend you start by purchasing Convertible RIs. They offer a smaller discount, but provide a trade-in capability which reduces the cost of errors. With experience under your belt, you can venture out to Standard RIs to save even more.
Finding Hidden Wastage
Since the introduction of “Instance Size Flexibility” feature in EC2, the probability of an RI going to waste has greatly dropped as the RI can automatically be applied to another sized instance in the same family. For example a reservation for a m4.large can cover 50% of the cost of a m4.xlarge (the other 50% is billed at on-demand rates).
To prevent wastage, a simple best practice is to standardize on one or two instance types. This greatly increases the probability that an unused reservation can be applied against an instance. In many cases, this is easier said than done since you may be relying on the instance size and family recommendations from the application vendors, which many not neatly line up to one or two instance types.
This means is that a wasted RI is sunk cost as there are absolutely no instances that it can be applied to.
HyperCloud 6.0 can analyze the cost and usage data gathered from AWS and find RIs that have not been applied and report these as wastage
If HyperCloud Analytics identifies a Standard RI lying unused, sell it on the RI Marketplace to recover cost before it becomes a sunk cost (Remeber Grampa Simpson from the begining of the blog?). If it’s a convertible instance, exchange it for one that was recommended in the “Potential Savings” dashboard.
Keep in mind that when you try to sell an unused RI on the RI Marketplace, AWS will charge you a fee based on the up-front price you paid. This means that even though purchasing RIs with all payment up-front is the best discount that you can get from AWS, these can be the most expensive ones to resell. This makes it even more important to use a predictive analytics tool like HyperCloud to help with your proactive RI selection.
Validating Your Decisions
What’s the point of saving money if you can’t show your team what a good job you’ve done?
HyperCloud 6.0 has a “Current Reserved Instance Savings” dashboard that can show you savings from current RI purchases. All reports can be downloaded as CSV files for further analysis or customization if required.
We believe our platform is beneficially different from any other in the market as it simplifies RI management with predictive analytics. Experience this for yourself by visiting us on the AWS Marketplace.