The question “How do I stop wasting money on Reserved Instances” often indicates the third stage of a business´s AWS “cloud awareness journey”. The first stage is the adoption of cloud computing services to take advantage of cost savings, flexibility and scalability. As the benefits of cloud computing are realized, companies expand their use of the service. This is the second stage. The third stage comes when companies realize they are spending more than they need on cloud computing services and implement measures to optimize their costs.
In order to prevent AWS costs spiraling out of control, many businesses reassign developers to write scheduling scripts. But is this the right approach? In some cases, writing and maintaining scheduling scripts could cost more than the business saves.
In the article below, ParkMyCloud´s founder – Dale Wickizer – shares his views on Reserved Instances; how they work, when they should be used, and when they should be avoided. Most importantly, Dale answers the question “How do I stop wasting money on Reserved Instances” with a viable and cost-effective option. Over to you Dale.
As Amazon Web Services (AWS) continues its explosive growth, customers find it increasingly challenging to control and optimize their AWS spending. Like any other business expense, if you don’t manage cloud spending, it can get out of control. As a result of dramatically increased adoption and consumption of AWS products and services, many customers have experienced sticker shock.
So how do you stop wasting money on Reserved Instances? First, a little background.
How do Reserved Instances purchasing options work?
The Reserved Instance purchasing option works as a 1- or 3-year contract and promises savings of more than 60%—the longer the commitment, the greater the cost savings in comparison to On-Demand. AWS also offers deeper discounts if you have more than $500,000 worth of Reserved Instances in a region.
However, longer commitment times involve risks:
- If AWS drops pricing, then the promised savings evaporate.
- When new instance types are introduced by AWS, these may attract people away from your contracts, which are based on the older types.
As a result, few customers make the 3-year commitment, and most settle for savings associated with the 1-year contract, which are in 30-x40% range.
There are also three different payment plans offered with Reserved Instances. Payment can be made either All Upfront, Partial Upfront + Monthly, or Monthly. The more you pay up front, the greater the savings. Take note, however, that if you opt for the 3-year term, then you must pay either All or Partial Upfront—the third payment plan is not available.
To some, the need to pay upfront and be locked in undermines both “pay as you go” and the notion of being “elastic”. It feels like a step backwards to the old economic model.
An example of the savings offered by Reserved Instances, along with what might happen if AWS cuts its On-Demand pricing by 30%, is shown below.
Where should I use Reserved Instances?
Reserved Instances are very much a “use it or lose it” proposition—in other words, there are no rollover minutes. Here’s why:
- The EC2 options available are specific to Region, Availability Zone, Instance Type (e.g. m4.large), Platform Type (e.g. Linux or Windows), and Tenancy. AWS, behind the scenes, attempts to randomly match instances you launch to the Reserved Instance contracts you have in place, based up these specific criteria. When there is a match, the cost benefit is applied. It is not uncommon for people to believe they are launching instances, which match all the criteria, when in fact they are not, so the contracts are under-utilized.
- AWS decrements the contract amount for every hour when not used, meaning your return on investment diminishes.
Given all of the tradeoffs mentioned above, Reserved Instances make the most sense in a production environment, where instances need to always be “on.”
Where should I avoid Reserved Instances?
In non-production environments, on the other hand, Reserved Instances are not your best bet. In fact, using ParkMyCloud to schedule on/off times for non-production EC2 instances is a much better option. The advantages are clear:
- Better Savings
- No Commitment or Upfront Payment
- Price Cut Protection
With ParkMyCloud, you can create parking schedules that automatically turn EC2 instances on and off according to specific needs. ParkMyCloud provides customized parking recommendations based on criteria specified by the user, which makes identifying “parkable” instances easier. The platform also tracks costs, projected 30-day savings, and actual savings for the current month.
For non-production instances, ParkMyCloud achieves EC2 savings of 50-73% with no annual commitment, upfront payment, or risk of instance termination or price cuts.
One of our newest clients cancelled a $10,000 order for AWS Reserved Instances in favor of EC2 instances he could turn on and off when he found out about our easy and powerful cost savings tool
However, if you have already purchased Reserved Instances and share them between production and non-production environments, not to worry! ParkMyCloud can be used to park the non-production systems, ensuring that production has more than their fair share of those contracts.
Overall, not too bad for a monthly subscription that starts at $29 per month!
You may also be interested our recent presentation, 5 Ways to Reduce Your AWS Spending.