AWS Cloud Cost Optimization

We hear questions like this on a weekly basis from CIOs who are getting a dose of reality as they watch their AWS bill go up every month like Amazon stock. IT leaders are initially excited about the potential benefits of their cloud migrations. The cloud offers increased scalability, performance, security, and flexibility--but it comes at a cost. If you manage that cost with skill and attention, the cloud can be significantly cheaper to operate than your legacy infrastructure. If neglected, however, high costs can cause you to pump the brakes on your cloud deployments in order to stay within budget.

Not managing AWS costs proactively is more harmful than just the dollars and cents you’re wasting--it costs you the opportunity to fully realize the benefits of the cloud in order to transform your business.

In fact, according to 451 Research, just over half of companies that recently moved to the cloud cite high costs as a significant pain point. These people were genuinely surprised by how quickly their bill got out of hand, because they thought the opposite would happen -- what a bitter pill to swallow! Companies choose AWS in the first place because of its unrivaled economies of scale and a history of consistent price cuts year after year, but they quickly realize that cloud costs don't manage themselves. You need a proactive strategy.

Luckily, cost optimization is a discipline you can master. Read on to learn what you can achieve when you have the right tools and the right strategies.

Before we dive into specific strategies, it’s important to understand why you would want to become proficient at AWS cost optimization in the first place. Obviously, AWS cost optimization will save you money; however, it’s more than that.

Over time, a poorly optimized AWS account can have non-obvious, ripple effects on your business. It becomes increasingly difficult to take advantage of the full capabilities of AWS because adding new services and capabilities often further increase your bill (which is already too high!). The end result is stagnation. And, with the pace of innovation increasing every day, and entire industries being disrupted by companies that embrace the cloud to move faster, stagnation is a death knell.

On the other hand, cost optimized organizations may be able to invest in new AWS services to unlock that next level of operational excellence. We've seen some companies reinvest their savings into new automation projects, allowing their engineering teams to work faster than ever. It can propel businesses to new heights!

It seems AWS releases a new service every day, and for those companies that have room in their budgets, they can be in a good position to take advantage of them quickly. For instance, your organization can be the first to use a new machine learning service to assist in making data-driven decisions.

The Bottom Line

As costs spiral out of control, bottom line results can be eroded; this can cause the business to react in a series of ways which are detrimental to its future. The money wasted has to come from somewhere, so companies may stop making other key investments such as hiring.

If you allocate costs by department (as many companies do), the department spending the most on AWS may be asked to cut back. Think what this could do if this same department is the most forward-thinking, innovative part of the business? Imagine what your company could do with the 30 percent savings from your AWS bill!

Unlocking Innovation

In businesses that have gone through a full digital transformation, people throughout the organization have access to game-changing capabilities only the cloud can enable, such as Artificial Intelligence, Machine Learning, and scalable compute resources which can solve previously unapproachable problems.

Once you control your AWS costs, you can inspire innovative use of technology throughout the organization--not just from within the IT department. Fully transformed businesses have cloud-savvy people dispersed throughout every department.

Speed and Scale

Marc Andreessen famously declared that “software is eating the world” almost a decade ago. His assessment is more true today than ever with software-powered innovations affecting every industry imaginable, and redefining winners and losers based on their ability to adapt to new software-driven business strategies. In a software-driven world, two facets of AWS infrastructure emerge as all-important: speed and scale.

Speed requires that you invest heavily in automation, which actually increases your AWS usage but more than offsets those costs with payroll savings. If you’re unoptimized, paying for additional infrastructure in AWS to enable that automation will be challenging.

Scale is where business models can be made or broken. With an inefficient infrastructure--for example, one that makes use of the wrong instance types and sizes--serving an additional million customers might actually lose the business a lot of money. As businesses scale out the infrastructure that supports their software-driven strategies, it’s key that they achieve the highest efficiency possible to reap the benefits from their success.

Sometimes, AWS Costs Go Unnoticed

According to Forbes, 70 percent of large enterprises in 2018 expected to increase their spending on the cloud by 20 percent or more--these are the known cloud costs.

But over the course of migration, it becomes clear for many that costs can go higher. Until they get the bill, though, it's easy to assume that costs will decrease simply because we expect them to. That expectation may be conditioned both by the very real promise of the cloud and by cost decreases AWS has made over time as efficiency improves.

For example, storage costs in AWS on a per-GB basis have declined 80 percent since the service was released in 2006, so it’s easy to assume your costs will naturally go down over time. Don't let this be you.

Understand your Bill

It may seem obvious, buy many customers miss the first step which is to truly understand where your costs are currently coming from. For many, it seems deciphering your AWS bill requires an advanced degree in forensic accounting and unless you are going stark raving mad, it can be difficult to understand what exactly you’re looking at.

Do you know which projects or departments are consuming the most resources? Is your usage increasing or decreasing? Who is in spinning up all of these resources and why? Can you map trends in your spending to other business indicators?

AWS Cost Explorer can be a good basic tool for making sense of your usage and it’s where most people should start. You can enable resource tagging within your AWS console to allocate costs to different departments or projects, and filter your view in Cost Explorer accordingly.

Clip Coupons

According to AWS, Reserved Instances (RI) can offer discounts of up to 75% when compared to On-Demand rates. It’s like a coupon: AWS matches your monthly usage against any RIs (i.e. coupons) you have and discounts the hourly rate if it finds a match! Unlike at the grocery store, RI matching happens automatically and doesn’t irritate the people behind you in line.

Seriously, understanding how to take advantage of RI’s is a crucial step to becoming cost optimized. The reason is twofold: first, RI’s provide savings for EC2 and RDS which are two of the largest components of most customers’ AWS bills. Second, the savings from RI’s can be so dramatic!

If you have workloads in AWS, paying on-demand rates for EC2 and RDS is like incinerating cash to stay warm. On-Demand is appropriate when you are unsure if a certain workload will persist for more than a year. However, when you consider that most workloads are migrating to the cloud and increasing in usage over time, it is safe to say you should be covering most of our workloads with RIs.

Many people worry about not being able to predict the size of instances they will need, but AWS has you covered. If you decide you need a larger or smaller instance type, you can modify your RI’s to retain the savings. For example, if you purchase an RI for a t2.medium and realize it’s overkill for what you need, you can split that RI into two t2.small RI’s!

AWS offers 2 different types of RI’s: convertible and standard. Here is a breakdown of the differences:

Characteristic	Standard	Convertible
Terms (avg.discount off On-Demand)	1yr (40%), 3yr (60%)	1yr (31%), 3yr (54%)
Change Availability Zone, Instance size (for Linux OS), networking type	Yes (Using ModifyReservedinstances API and console)	Yes (Using ExchangeReservedinstances API and console)
Change instance families, operating system, tenancy, and payment option		Yes
Benefit from Price Reductions		Yes
Sellable on the Reserved instance Marketplace	Yes (After linking account with a US bank account)	Coming soon

RI’s give you a bigger savings the longer you commit, and the more you pay up front. The smallest savings come from a 1-year no upfront RI, and the largest from a 3-year all upfront RI.

Right-size

When you provision servers on-premise, you have to make sure you have enough capacity for today, plus enough capacity to accommodate the next few years of growth because re-provisioning physical servers is such a pain.

It would make no sense to buy a physical server that just barely met your needs because you would be buying a new one in no time, and re-racking and re-stacking. In the cloud, re-sizing a server can be done with the click of a mouse, so it makes a lot more sense to right-size your instances to just above you need.

Right-sizing not only means adjusting your server size, it also means adjusting your instance type. AWS offers general purpose instances that offer a good balance of performance. but they also offer instance types that are optimized for workloads for memory, disk, graphics, or CPU performance.

Understanding whether your workloads are balanced or would benefit from a specialized instance type, can 50% savings or more when compared to being in the wrong instance type. If customers are unsure, we often tell them to start with M-class instance and look for the bottleneck by reviewing performance metrics.

Compliance

Duis autem vel eum iriure dolor in hendrerit in vulputate velit esse molestie consequat, vel illum dolore eu feugiat nulla facilisis at vero eros et accumsan et iusto odio dignissim qui blandit praesent luptatum zzril delenit augue duis dolore te feugait nulla facilisi. Lorem ipsum dolor sit amet, consectetuer adipiscing elit, sed diam nonummy nibh euismod tincidunt ut laoreet dolore magna aliquam erat volutpat.

Join the New Generation

When AWS releases a new generation of instances based on the latest advances in technology, the price/performance of the new instance type is always better than the old one. For example M5 instances are 14% more efficient than M4’s and up to twice as efficient for certain workloads.

Many customers launch a particular instance and just let it hum along for years. When a new instance generation is released, we recommend you evaluate it and deploy it as quickly as you can. For many customers, they can upgrade and downsize at the same time without any loss of performance.

Don’t Pay for Unused Resources

Developers and systems administrators are brilliant and wonderful, but they sometimes need help cleaning up after themselves. Recently, we had a customer whose developer had attached an extremely high performance disk to a server that was suffering from poor performance.

The problem ended up having nothing to do with the disk drive, but the developer forgot to destroy the unnecessary disk and incurred a $7000/mo charge on their bill until they noticed it and destroyed the volume. That’s like forgetting you had leased a Lamborghini and just leaving the car at the dealership. Even smaller resources like unused Application Load Balancers (ALB), Amazon Machine Images (AMI), and Elastic Network Interfaces (ENI) can really add up!

As you can imagine, you have to can’t just check in on these issues once and be done with it. You have to keep doing it over and over again on a regular basis to stay on top of the changes.

Modernize your Infrastructure

Perhaps the most often overlooked way to reduce costs in the cloud is modernize your infrastructure. Many customers just migrate their infrastructure to the cloud like-for-like--a “lift and shift” migration as it’s known. Creating a datacenter in the cloud is an enormous missed opportunity.

By re-architecting your infrastructure to take advantage of modern design patterns, you can replace servers with scalable, fully managed services that cost a fraction of the servers they replace. For example, if you previously operated a load balancer appliance in your infrastructure, replace that with an Application Load Balancer which may cost as little as $0.02 per hour to operate. If you have an SMTP server in your stable of servers, replace that with Amazon’s Simple Email Service which is free for the first 62,000 emails you send per month.

Join a Buyer’s Club

In AWS, some services have tiered rates based on usage volume. Examples include S3, CloudFront, and Data Transfer. Unless you have massive usage across all regions you are in to achieve the lowest pricing tiers, you are leaving money on the table. By grouping your usage with other organizations, you can get a blended rate that is far lower than you would by pay on your own.

Technically, you could try to get some other companies together on your own and try to split your bill equitably and do it all yourself, but that’s complete madness. The best approach to get lower rates is to join a managed billing service like Privo’s AWS Managed Billing Service (which is free). Other AWS consulting firms might offer it for a small fee or as part of a larger managed services engagement.

Not all managed billing services are created equal, however. Make sure your provider has dedicated billing specialists on staff that have the depth of knowledge to navigate this complicated topic expertly!

If you need a real life example, consider Promosis, a company that uses gamification strategies such as fun sweepstakes and skills contests to help clients build relationships with their customers. Promosis enthusiastically embraced the cloud and very rapidly adopted it to power their platform. Over time, this meant increasing costs that became difficult to manage. Manually responding to cost fluctuations was no longer practical, since costs changed dynamically alongside user demand for the Promosis platform.

This meant Promosis needed a new strategy for managing costs. They needed a solution that could track and respond to real-time adjustments in costs. Privo's cost optimization portal ultimately provided the answer they needed by allowing Promosis to view detailed AWS usage and cost cutting recommendations.

The result?

Here's what happened:

30%

Reduction in AWS hosting costs

40%

Savings in database management costs by migrating to Aurora

5X

Increase in database performance

Right-Sizing

With the correct resource usage levels, Promosis found they could save money while utilizing their resources more effectively. They no longer over or under-use their AWS resources.

Reserved Instances

Promosis could see various reserved instances options that enable various return on investment (ROI) values. Privo's portal quantified the value Promosis could gain from these opportunities.

Proactive Cost Management

Since Promosis has their AWS usage alongside practical recommendations right in front of them, it is now easier to plan ahead and manage ongoing costs. The Promosis team can make adjustments and understand usage without having to worry about being surprised by an unexpected bill.