If you’re using AWS, you’ll already be well aware of the cost-cutting potential of utilizing cloud services.
But sometimes, just adopting the cloud isn’t enough to really make a dent in your bottom line.
Just like any other kind of outlay, you need to keep a watchful eye over your cloud environment to make sure you’re not racking up any extraneous costs or paying for more resource than you need.
To help you keep your cloud spending in check, we asked five AWS experts for their top tips on keeping cloud costs down and efficiency up.
Meet our AWS experts
Aaron Rees is a cloud innovator and co-founder of Rebura, one of the UK’s fastest-growing AWS consultancies.
Ewere Diagboya is Africa’s first AWS Community Hero, a tutor, and a Cloud Evangelist at nOps and nClouds.
Julio Faerman is a Software Engineer at Nubego and an AWS Technical Evangelist, who helps builders be successful with AWS.
Marc Weaver is a certified AWS Solutions Architect and founder of databasable, a cloud computing consultancy that specializes in AWS.
Trevor Sullivan is a founder, engineer, instructor, consultant, and Solutions Architect who currently holds seven AWS certifications.
Efi Merdler-Kravitz is Director of Engineering at Lumigo, a monitoring and debugging platform for AWS serverless apps.
Go serverless
Serverless computing is a method of executing code in which all the resources you need to run an application are handled dynamically by the cloud provider.
These backend services are provided on an as-used basis, and you’re only charged for the amount of resource used to run the code.
It’s used by the likes of Thomas Reuters and Netflix to enable them to innovate quickly and ensure high availability.
Like hailing a taxi when you need to go somewhere, rather than paying to run a car that you don’t need to use all of the time, with serverless computing, you only pay for the time your code is actually being executed.
That means no money wasted on idling resources.
You can use serverless for a tonne of different things, not only compute power for developing applications, but also for the likes of storage, databasing, APIs, and analytics.
When you use AWS Lambda, Amazon’s serverless computing service, your first million requests per month are free, and every million after that cost $0.2 a pop.
With AWS handling all server infrastructure, security, and scaling for you, there are all sorts of longer-term savings to be made too.
There are a number of case studies out there in which companies have slashed their AWS bill dramatically by utilizing serverless, and it can be implemented more quickly than you might think.
“There are two types of cost-cutting. The first option is to stop using your own managed servers—keeping the servers operational is expensive, especially in times when the typical workload decreases—and move to managed services like RDS or Amazon MQ. You can do this right now.
“But, for these types of services, you’re still paying for the unused workload; you can quite easily move to a serverless counterpart without a revolutionary architectural change and pay only for usage.
“So the second option is to stop using managed services and move to serverless solutions like DynamoDB, Serverless Aurora, or SQS, where you pay only for usage. That way, you get all the benefits of the first option and cheaper operational costs.
“Especially now that workloads have become unpredictable, serverless solutions can reduce costs. You only pay for what you use. Even if you have “monolithic” applications, you can move them to serverless without too much work; although they won’t be optimized, they will save you money.
“Solutions like AWS Serverless Express for NodeJS-based applications or Zappa for Python-based applications can quickly move you to the serverless world with minimal hassle.”
Efi Merdler-Kravitz
More AWS talent than anyone else
Take a look at our database of pre-screened AWS professionals and take the first step toward landing the best administrators, developers, and consultants in the market.
Take a lookDon’t overprovision
“EC2 instances are the most ripe for over-provisioning. It’s easy to spin up a virtual machine that has much more capacity than your applications actually need.
“To solve this, you can “rightsize” your EC2 instances by examining built-in CloudWatch metrics and determining if the CPU, memory, disk, and network resources are being utilized heavily.
“Based on your observation of these metrics, you can decrease the resize of the instance to a smaller instance type until resource utilization increases to a normal level, roughly 50-80%.”
Trevor Sullivan
AWS provides recommendations that can help you get started with rightsizing. Open up your AWS Management Console, go to the Billing and Cost Management section, and select Cost Explorer.
Once you’ve launched the Cost Explorer, it’ll generate a list of recommendations.
These recommendations will include a breakdown of EC2 instances that are being used minimally or not at all, and make suggestions for rightsizing them.
You can then choose to enable these recommendations within the console.
These recommendations can be filtered and examined by a number of KPIs:
- Optimization opportunities, which is the number of recommendations the console currently suggests actioning based on your current usage
- Estimated monthly savings, which is the total amount of monthly savings you could make by implementing existing suggestions
- Estimated savings, which is your estimated savings relative to the direct instance costs that would be incurred by actioning the recommendations in the current list
If you want to go a step further in making sure you’re always rocking the right sizes, you can automate the scaling of your instances.
AWS Operations Conductor will automatically resize your EC2 instances based on the recommendations generated by Cost Explorer, so you don’t have to dip into the report on the regular to check whether you’re still running the right size instances.
Terminate instances you’re not using
Not only can you save by making sure you’re using the rightsized instances, but you can also cut costs by using them at the right times.
To optimize your spending, it’s crucial that you shut down instances when you aren’t using them, rather than leaving them to tick over.
There are tools to help you achieve this, like AWS Instance Scheduler. This service allows you to set up custom start and stop schedules for your EC2 and RDS instances.
According to AWS, if you use this tool to shut down your instances outside of regular business hours, you could save up to 70% compared to running them around the clock.
Looking for top AWS talent? We make it easy.
Tell us what you’re looking for and we’ll put together a job spec that’ll attract professionals with the skills and experience you need.
Share your specReserve your capacity in advance
One of the great benefits of using cloud services is that everything is on-demand; you use what you need, when you need it, without having to plan out in advance.
That said, there’ll probably be at least a good proportion of your resource that you’ll always need as a baseline.
If you know that you’ll likely use a certain capacity of EC2 compute power, you can make significant savings by reserving that capacity and committing to using it over a 12 of 36 month period.
This is where those AWS Cost Explorer recommendations can come in handy once again.
These reports will show you how much resource you’re using within services like EC2, RDS, Redshift, and ElastiCache, and will even recommend which instances to reserve to save the most money.
Buying a Reserved Instance could net you savings of up to 70%, but you will have to pay for it regardless of how much capacity you end up using.
There are three types of Amazon EC2 RI to choose from.
- Standard — These RIs provide the most significant potential discount, priced around 72% lower than on-demand instances. You can also sell your reserved capacity on AWS Marketplace if you don’t use it, but you can’t convert them into another instance type should your needs change.
- Convertible — More expensive (around 54% discount), and you can’t sell them on, but you can change the instance attributes if necessary. It should be noted, though, that if you scale up your convertible instance, you can’t downgrade it again later, so start small!
- Scheduled — Available to launch within set, reserved windows, Scheduled instances mean you can reserve capacity for when you know you’re going to have a short spike in capacity demand.
You can pay for RIs fully upfront, partially upfront, or every month.
“Maximizing return on your cloud investment is an area I’m really passionate about. It’s amazing some of the savings I’ve seen customers make.
“There’s a whole variety of ways to do this with AWS, like Reserved Instances and Savings Plans, but you’ve got to understand your environment and make the right decisions.
“Before making any Reservations or adopting Savings Plans, I’d always recommend a thorough investigation of your system, followed by a rightsizing exercise to make sure you’re reserving the right capacity.
“Again, AWS is pay-as-you-go, so there’s no need to overprovision and overpay!”
Aaron Rees
Take advantage of freebies
As well as its substantial stable of “always free” products, there are many AWS services you can use free for a year.
If you’re looking for ways to shave a little off your bill, you could always take advantage of one of these offers to trial whether or not switching to another service or trialing a different type of infrastructure could be cheaper before you commit.
Sometimes you can catch a special offer too.
Amazon WorkSpaces is a fully managed Desktop-as-a-Service solution that lets organizations roll out fast, responsive, and secure virtual Windows or Linux desktops that users can access from any device.
You can get WorkSpaces for up to 50 users now, up until June 30, 2020.
There are also programs that you can apply to if you’re a certain type of business. AWS Activate, for example, is a free scheme aimed at giving startups and early-stage entrepreneurs credits to help them get started on AWS.
Whether they’re new to the cloud or getting ready to scale up, AWS Activate offers up to $100,000 worth of AWS bucks, plus tech support and training to eligible businesses.
Implement machine learning tools to forecast spending
“Migrating and optimizing cloud resources is a known and clear way to cut costs. But deeper than that, the cloud enables innovation towards making things simpler and more efficient.
“Adjusting instance types and avoiding waste surely helps to save money, but access to great technology, like machine learning-based forecasting, can change cost structures and curves entirely.”
Julio Faerman
[/su_column] [/su_row]It’s a longer-term strategy than some others mentioned here, but an essential one.
Tools like Amazon Forecast can use your historical data to predict demand and forecast business outcomes.
Amazon uses this same technology to help get ahead of product popularity, and resource needs to make sure it’s always ready and able to meet customer requirements.
Having this kind of accurate data to hand means you can better plan finances and resources, cutting down wasted outlay and helping to maximize revenue.
Practical advice from AWS experts to help you weather the storm
With extensive insights, advice, and best practices from cloud leaders, our brand new white paper is the ultimate guide to optimizing your business with AWS.
