Optimizing the AWS Usage

Michael Meyer, Chief Risk Officer (CRO) and Chief Security Officer (CSO), MRS BPO, LLC

Almost all of us use AWS in one way or another these days. However, many of us signed up for the free trial, started using their services and never gave it another thought because we were saving so much money over the on-premises equivalent. You know what, that ‘set it and forget it’ mentality is now costing you a lot of money and fueling the growing profits of AWS. 

So, what can you do to take advantage of savings from 30 percent to 90 percent off while using AWS?  Yes, you read that right 30 percent to 90 percent off!  No, youdon’t need to redesign everything, switch to another cloud service or even use another company such as a broker to get this lower pricing.  You can, however, get this lower pricing with a simple change, and all you need to know is just one thing—your average usage over time.  Yeah, I know this sounds pretty simple, but for some of us, understanding our usage patterns might take some homework. 

We will cover the most common situation, which is that you are already using AWS, and that as time goes on you have moved more of your workload to AWS, causing the overall cost to increase.  You need to find a way to lower these costs and soon.  The dramatic savings potential is realized through the different pricing options and features that AWS offers.  The funny thing is that most of us haven’t really looked into all of the different ways that AWS prices their commercial services, but we certainly know all about the Amazon consumer side and when prime day is. 

  It's like filling your gas tank to save money when returning a rental car instead of them charging you double to do it​  

At a high level, AWS has four paid usage pricing options. These pricing options are On-Demand, Spot Instances, Reserved Instances, and Dedicated Hosts. These options come with billing per hour and per second. Most of us use the first pricing option, On-Demand, which is the easiest and most flexible to use.  This option means that we don’t have to think about, plan or know anything about our usage in advance.  Unfortunately, this option is also by far—the most expensive one.

The next option is Spot Instances (Spot).  This is a really cool option where AWS offers you the ability to buy spare AWS computing power during off-peak times. This means that if you can wait and schedule your applications to run maybe even just a few minutes later, you can achieve the most significant savings of all options—up to 90 percent.  It's like filling your gas tank to save money when returning a rental car instead of them charging you double to do it.  This lower cost also has a tremendous upside that might be critically important to your operations, and that is you can get up to 9X more capacity for the same cost that you are paying now.  Said another way, by using Spot you might be able to run something up to 9X faster for no additional cost! This isn’t hard to manage either because AWS created some neat new tools called Spot Instance Advisor and Spot Fleet that can automatically schedule and manage this capability for you. The only downside to Spot is that if AWS runs out of capacity and needs it for a higher paying customer, AWS will automatically pause your instance (with a few minutes notice) until it has capacity again and then automatically resume your instance. AWS even thought of a way around that by allowing you to schedule dedicated Spot time for up to six hours, though you give up some of the savings and only save 30-50 percent off.   Overall using Spot is an entirely hands-off experience that saves tons of money.

The third option is Reserve Instances. These can save up to 75 percent because you buy in bulk what you need in advance. It is like going to BJ’s or Costco instead of a local supermarket, and as a reward, you get wholesale vs. retail pricing. Since AWS can plan for and predict how much capacity they need in advance, they pass on the cost savings to you. The downside here is that it means you will need to know how much usage your services will have over a month’s time. For most of us, it might be a little tricky which is why AWS provides numerous usage and various metrics reports. Since the cost savings is so high, even if you reserve a lower amount than what you use now, you are still way ahead of the game, plus you can adjust your reserve capacity, so mistakes can easily be minimized.  

The last option is called Dedicated hosts.  These are probably the hardest option to use because these are dedicated servers and require you to install your company’s licenses on them to be used. While you save up to 70 percent by using this option, you still must own and manage the various software licenses. 

Of course, there are many other ways of saving money with AWS by optimizing your usage, moving to a different region and looking more granularly into how you use each service.  To help with this, AWS has a cost optimization tool called AWS Simple Monthly Calculator.  The Simple Monthly Calculator helps you estimate in-depth what your monthly bill will be, so you can see how to save the most money while using AWS. They also have a pretty neat Cost Explorer tool that has a number of default reports that will help you take a deeper look into your usage and allows you to group the costs which makes it easier to map back to specific cost centers too. For more information take a look at their “How AWS Pricing Works” Whitepaper which has a lot of specific service pricing details. Once you get smarter about how you use AWS and get hooked on how to save more or get increased runtime for less, you won’t go back to the old ways of paying retail pricing for AWS.