Over the last decade, the cost of Amazon Web Services (AWS) has become a primary concern of businesses. That's no surprise: AWS has many tools that offer a range of IT resources – from IT infrastructure and bandwidth to analytics tools and SaaS applications – and each affects the total cloud bill. Understanding and managing cloud costs isn't simple with AWS.
A report from research firm Enterprise Management Associates found that organizations struggle with cloud issues like unused capacity, overuse, and suboptimal application placement. The result is unpredictable bills and, in the end, wasted money. In fact, EMA found that a typical organization burns through about 30 percent of its cloud budget for unneeded resources. Given the size of many company's AWS bills, this is a significant amount of money. So businesses feel real urgency to reduce cloud costs.
Here are seven strategies your organization can use to lower its AWS bill and get its cloud costs under control, focusing on using the Amazon cloud platform optimally.
- Use the AWS billing report
- Use outside analytics tools
- Get discounts and 'volume pricing'
- Review your infrastructure
- Use the AWS spot market
- Consolidate cloud vendors
- Consult technical experts
The complexity of cloud pricing models is an ongoing challenge. To address this, AWS offers its own billing dashboard. It provides a high-level view of costs, including services such as EBS, Amazon S3 and Amazon EC2. However, for many organizations, this tool does not deliver detailed data that leads to insights.
One option is to use the billing report Amazon offers. It lets you download a zipped .csv file with each billable item on its own line. Together with custom tags, it’s possible to import the data into an analytics tool and group costs by features, departments, projects, or whatever other criteria your organization requires. AWS also offers Trusted Advisor and Cost Explorer apps that deliver additional insights and helps spot cost trends.
Not surprisingly, there are a number of third-party tools designed to peer into AWS data. These include SaaS applications such as Cloudyn, Cloudhealth and Teevity, which deliver sophisticated functionality. Also available is the open source resource Ice, which can sort AWS data into distinct categories, such as accounts, regions, service types (EC2, S3, EBS), usage types (EC2 - m1.xlarge, for instance) along with custom cost and usage categories. Many of these tools use APIs to connect disparate data sources more effectively.
A big challenge for organizations in today’s instant-on environment is overseeing developers and teams that enter a credit card number and turn on a cloud service. In the rush to get compute capacity or applications up and running, they may not check to see whether the enterprise has a volume pricing discount in place.
In other cases, an enterprise may have started with only a few AWS services but, over time, added additional components and services. The result is pricing that’s higher than what the organization could realize by pooling all the services and negotiating a lower overall price based on volume. Organizations might also qualify for specific discounts and bonuses. But unless you know your data and service usage – or ask for a review from an AWS sales rep – you wind up paying higher prices.
As organizations grow and change, infrastructure accumulates. It’s no different with cloud services such as AWS. As a result, it’s important to periodically review the cloud services and internal architecture your organization uses and ensure that you’re not buying things you don’t need. In many cases, this task is best accomplished using a cloud team or task force. Technical factors that contribute to wasted dollars typically revolve around four primary factors:
- Detached Elastic IPs (EIPs) that come at no additional cost with an EC2 instance but ring up a bill when they are not properly categorized.
- Block stores that accumulate and become detached from root EBS volumes. Most organizations do not reattach these block stores and wind up paying more as a result.
- Load balancers that manage traffic without being monitored. When resources are switched off, an organization can find itself paying for load balancing on these phantom resources.
- Inactive instances that haven’t been used for a week or more. There may be a good reason to keep these instances in place but, more often than not, they represent resources that cost money but provide no service or value.
Part of the appeal of the cloud is the high level of flexibility and scalability it offers. While there’s a clear benefit of using defined AWS resources for various services and tasks, there’s also an opportunity to put the spot market to use – particularly for non-critical tasks.
Using auction pricing, it’s possible to obtain discounts or rebates as steep as 90 percent. You simply place a bit and, if it is accepted, you receive one hour of the instance. If your organization receives a termination or reclaim notice – meaning that the market price has risen above the price you paid – you have two minutes to terminate. This gives you time to save the state, upload any final files, and exit.
It’s often necessary to rely on multiple vendors – such as Google Cloud and Microsoft Azure along with AWS – to address a spectrum of cloud computing requirements. Of course, many smaller, niche vendors might also enter the equation – particularly as marketing, finance, human resources and others pull the strings on essential needs. In fact, a RightScale report noted that the average company uses 4.8 major cloud providers.
However, along the way, strategic decision-making can become chaotic when groups, departments and divisions fail to operate in a coordinated way. What’s more, over time, the sum of clouds services can lead to significantly higher costs.
The upshot? Organizations can benefit by reviewing services and vendors and then work to redistribute and realign cloud procurement and cloud use in a more optimal way. It’s also possible to benefit by consolidating providers. This requires a detailed review of clouds and a plan for strategically realigning them. It also involves a review of pricing and a clear understanding of how services work before committing to them.
It’s not uncommon to overprovision or overscale AWS cloud resources. Although autoscaling tools are valuable – and they typically result in savings – they can also wind up on autopilot and lead to wasted capacity and resources.
In addition, an enterprise may be using clouds for storage or other tasks when conventional tools and solutions offer a more cost-effective approach. Still another possibility is the use of serverless services, containers and automation solutions. All of these – as well as the combination of tactics – can dramatically lower costs for AWS as well as other cloud computing frameworks.
The key is to give technical experts some time and resources to review and analyze your AWS cloud framework. In most cases, they can make suggestions about how to trim costs and, at the same time, strategically improve the overall cloud strategy.
To be sure, there’s no single route to a more cost-efficient AWS infrastructure. Savvy organizations understand that numerous small gains can lead to big savings – while ensuring the highest level of cloud services.