The Complete Guide to AWS Compute Services
Welcome to the world of AWS compute services! It’s a big one, but compute really only scratches the surface of all the resources AWS offers. In this article, we’ll walk you through each and every AWS compute service, explaining their features and pricing structures. Let’s dive in.
Table of contents
- What are AWS compute services?
- Comparison of AWS compute services
- List of AWS compute services
- Amazon Elastic Compute Cloud (EC2)
- Amazon EC2 Auto Scaling
- Amazon Lightsail
- AWS App Runner
- AWS Batch
- AWS Elastic Beanstalk
- AWS Fargate
- AWS Lambda
- AWS Local Zones
- AWS Outposts
- AWS Serverless Application Repository
- AWS SimSpace Weaver
- AWS Snow Family
- AWS Wavelength
- VMware Cloud on AWS
- Pricing for AWS compute services
- Get your head in the clouds
What are AWS compute services?
AWS compute services are a line of infrastructure as a service (IaaS) products from Amazon Web Services (AWS) that allow customers to purchase computing power as needed instead of maintaining their own physical servers.
The AWS compute platform includes several different categories, such as:
- Instances (virtual machines)
- Hybrid and edge locations
- Cost and capacity management
Following the release of Amazon Simple Storage Service (S3) and Simple Queue Service (SQS) in 2006, Amazon Elastic Compute Cloud (EC2) was a foundational AWS product and the company’s first IaaS compute product. Compute is one of the most popular AWS services today.
Comparison of AWS compute services
AWS has quite a few products on offer, so it can be easy to get them mixed up. Here’s a brief rundown of every AWS compute service at the time of publication.
|Instances||Amazon Elastic Compute Cloud (EC2)||Lets you rent and manage virtual servers in the cloud with elastic compute power. Uses an on-demand pricing model.|
|Amazon Lightsail||Lets you quickly launch resources for small projects. Works best for preconfigured apps, like setting up a WordPress site.|
|Amazon Batch||Lets you process large workloads in smaller chunks, or batches.|
|Containers||Amazon Elastic Container Service (ECS)||Container orchestration tool that lets you deploy and manage Docker containers.|
|Amazon ECS Anywhere||Lets you run self-managed containers on your own infrastructure.|
|Amazon Elastic Container Registry (ECR)||Lets you deploy container images to Amazon ECS, EKS, or on-premises.|
|Amazon Elastic Kubernetes Service (EKS)||Lets you run Kubernetes applications and clusters in AWS cloud or on-premises.|
|Amazon EKS Anywhere||Lets you run Kubernetes apps on your own infrastructure, similar to how ECS Anywhere lets you run self-managed containers.|
|AWS Fargate||Lets you deploy and manage Docker containers using a serverless compute engine|
|AWS App Runner||Lets you run containerized web apps without having to configure orchestrations, manage servers, or build pipelines.|
|Serverless||AWS Lambda||A serverless compute service that lets you run code without having to manage servers or clusters.|
|Edge and hybrid||AWS Outposts||Lets you take advantage of hybrid cloud by delivering AWS infrastructure and services to your on-premises or edge locations.|
|AWS Snow Family||Lets you securely move data in and out of AWS through several different hardware options, as well as operate edge locations|
|AWS Wavelength||Lets you take advantage of mobile edge computing infrastructure for fast applications on 5G devices|
|VMware Cloud on AWS||Lets you run VMware workloads on AWS infrastructure through a managed service|
|AWS Local Zones||Lets you run workloads in closer physical proximity to end users for latency-sensitive applications|
|Cost and capacity management||AWS Savings Plan||Lets you save up to 72 percent on compute services through a flexible pricing model|
|AWS Compute Optimizer||Lets you improve improve price performance and optimize costs through automated recommendations from AWS|
|AWS Elastic Beanstalk||Lets you easily deploy applications by automatically managing the AWS services you need to host and scale your application.|
|EC2 Image Builder||Lets you easily and quickly build and maintain secure Linux or Windows Server images.|
|Elastic Load Balancing (ELB)||Lets you improve availability by automatically distributing traffic across one or more Availability Zones (AZs).|
List of AWS compute services
Amazon Elastic Computer Cloud (EC2)
One of the most popular products AWS offers, EC2 allows you to rent and manage virtual servers in the AWS cloud. The “elastic” in EC2 comes from the product’s flexibility in terms of customers’ ability to create, launch, and terminate EC2 instances as needed.
- Ease of deployment—you can launch applications directly to EC2 instances at the click of a button
- Affordability—deploying instances on virtual servers running on physical servers means lower cost of compute, and EC2 offers a free tier that includes 750 compute hours per month
- High availability and low latency—you can deploy EC2 instances across multiple locations, which include Regions and Availability Zones (AZs)
- On-demand. Pay for compute by the hour or by the second.
- Spot instances. Lets you take advantage of unused EC2 capacity across the AWS cloud for as little as 10 percent of the cost of on-demand.
- Reserved instances. Pay a discounted rate to reserve a specific instance type in a certain Region for a term of one or three years.
- Dedicated host. Pay to rent physical servers totally dedicated to running your instances.
- Savings plans. EC2 offers various volume discounts when you commit to a certain amount of compute usage for a term of one or three years.
Amazon EC2 Auto Scaling
This isn’t actually a separate service but rather an option within EC2 to enable autoscaling on your instances. You can set scaling policies to control when your instances will automatically scale, and EC2 Auto Scaling offers dynamic and predictive scaling policies that use artificial intelligence to autoscale.
- No additional cost. Amazon EC2 Auto Scaling is included for whichever pricing plan you select for EC2.
Best for small projects, Amazon Lightsail lets you quickly launch resources. Lightsail is intended to be used by people with little to no AWS experience and works great for launching preconfigured apps like WordPress sites.
- Intuitive, easy-to-use graphical user interface that’s great for beginners
- Includes virtual private server (VPS) instances, SSD-based storage, containers, DNS management, a static IP address, and more
- Good for small business applications, building test environments, and dipping your toes into the AWS cloud before moving on to more robust products like EC2
- One monthly price. You’ll need to contact AWS for a pricing quote, but you can get started with Lightsail at one monthly price for as low as $3.50
- Free tier. Lightsail is part of the AWS Free Tier, and AWS sometimes runs promotions
AWS App Runner
With AWS App Runner, you can run containerized apps and APIs without having to worry about managing the more technical infrastructure aspects, like configuring orchestrations and building pipelines. This makes App Runner a fully managed service from AWS, making it a good choice if you want to build an application or website but you don’t have past infrastructure or container experience.
- Automatic deployments from your source code or container image
- Automatic load balancing
- Auto scaling
- Fully managed TLS certificate management
- On-demand. Pricing for App Runner is based on the compute (vCPU) and memory resources of your container instances. You only pay for compute resources when your container instances are running. You also pay for the memory your idle apps use for any provisioned container instances.
- Monthly fee. App Runner charges for all automated deployments you make in a month. Automatic deployments cost $1 per application, per month.
- Build fee. If App Runner has to build your container from your source code, it will bill you for the time it takes to do so. The build fee is $0.005 per build-minute.
AWS Batch lets you run large quantities of batch and ML computing workloads in smaller chunks called batches. The ability this gives customers to run hundreds of thousands of small processing jobs makes Batch a powerful tool for the pharmaceutical, semiconductor, and life science industries.
- Dynamically provisions instances based on volume
- Lets you run batch jobs on EKS clusters
- Works with Fargate for a totally serverless architecture
- Priority-based job scheduling
- No additional cost. Batch is included when you pay for other AWS resources like EC2 or Fargate.
AWS Elastic Beanstalk
Another managed service, AWS Elastic Beanstalk lets you deploy web apps without having to fuss over provisioning and managing infrastructure. Use the Elastic Beanstalk console to answer questions that tell AWS how to create your cloud environment. Add your code, and Elastic Beanstalk will handle the runtime to deploy your app with the right mix of AWS cloud resources.
- Supports a number of programming languages, like Java, .NET, Go, and Python
- Comes with Elastic Load Balancing and Auto Scaling out of the box
- ISO, PCI, SOC 1, SOC 2, and SOC 3 compliant
- No additional cost. Elastic Beanstalk builds your cloud environment, and you pay for the resources it uses as they are normally billed.
AWS Fargate is a serverless compute engine for deploying and maintaining containers. Fargate supports Docker containers, and it’s compatible with Amazon Elastic Container Service (ECS) and Amazon ELastic Kubernetes Service (EKS).
- Includes automatic scaling
- Works with other Amazon container services like ECS and EKS
- On-demand. Pay only for what you use. That includes vCPU, memory, operating systems, CPU architecture, and storage resources.
- Spot pricing. Fargate Spot lets you take advantage of spare capacity at up 70 percent off the on-demand price.
- Ephemeral storage pricing. All Fargate Tasks and Pods come with 20GB of ephemeral storage.
- Savings Plans. Commit to use a specific amount of Fargate compute for up to 50 percent off the normal price.
AWS Lambda is a serverless compute service, meaning you can write code (called functions) without having to worry about managing servers. Lambda supports many popular coding languages like Go, PowerShell, Java, Node.js, C#, and Python, and it scales automatically.
- Lets you add custom logic to AWS resources
- Lambda can execute your code in response to different events
- Lambda functions do have a 15-minute timeout, so it’s not ideal for processes that last any longer than that
- Supports code authorship using the IDE of your choice or via the Lambda console
- On-demand. Only pay for what you use—there’s no charge if your code isn’t running. Pricing is based on the number of requests for your functions and the time it takes to execute your code. Note that testing counts as a billable request.
- Free tier. Included in the AWS Free Tier, which includes one million free requests per month
AWS Local Zones
AWS Local Zones is a solution for running applications with low-latency requirements by running them closer to end users. This lets you take advantage of edge computing, making it a popular choice for makers of online games, AR/VR, and video live streaming.
- Supports a variety of AWS cloud resources like EC2, EKS, ECS, EBS, RDS, ElastiCache, and more for multiple locations like Atlanta, Chicago, LA, Hamburg, Perth, Taipei, and more
- Resource- and location-based. Pricing for Local Zones depends on the AWS resources you use and where you use them. Due to the highly specific nature of edge computing, you’ll have to check local rates for the services you want to use. To make this easier, contact AWS for a quote.
If you want access to AWS resources but you need to keep your workloads on-premises, AWS Outposts is your solution. Outposts let you run AWS resources in your internal data centers, making it a good option for organizations that use a hybrid cloud environment. What’s more, AWS actually handles delivery and installation of Outpost servers in your data center themselves.
- You can choose between AWS Outposts racks and servers. A rack is a 42U form factor, while a server can be a 1U or 2U form factor.
- Run AWS resources in your own data centers.
- Integrated networking gear for Outpost racks
- Access to a number of compute services, including EC2, ECS, and Sagemaker Edge Maker. More options for racks.
- Monthly pricing. Prices vary based on your location and the services you use, but all services come at a monthly price paid over a three-year term.
AWS Serverless Application Repository
The AWS Serverless Application Repository isn’t a service but a managed repository of pre-built, reusable serverless applications. If you use any serverless architectures, this is a great resource to explore for saving time and labor in building new applications from scratch.
- Compatible with AWS Serverless Application Model (SAM)
- Use pre-built applications without having to clone, build, package, or publish source code to AWS
- No additional cost. The best part about Serverless Application Repository is that it’s free to use. You’ll still pay for the AWS cloud resources you need to deploy, run, and maintain any applications you use, but use of the repository itself is free.
AWS SimSpace Weaver
AWS SimSpace Weaver is a powerful service for making 3D simulations. One of the reasons SimSpace Weaver is so powerful is because it’s fully managed—meaning you download the SDK, integrate your code using built-in APIs, and upload your application. SimSpace Weaver takes care of deploying and managing your infrastructure.
- Fully managed infrastructure frees you up to focus on building your simulation
- Spatial partitioning lets you break up your simulation world into distinct areas managed by its own SimSpace Weaver app
- Ability to run your simulation code locally for free testing before you’re ready to move to the cloud for production
- Built-in integrations with Unreal Engine 5 and Unity LTS Release 2021.3.7f1
- No additional cost. There’s no extra cost to use SimSpace Weaver. Rather, since this service runs simulations on EC2 instances, you pay an hourly rate for each EC2 instance you use.
AWS Snow Family
AWS Snow Family isn’t a service but a line of hardware products. If you’re an enterprise organization needing to move a lot of data—as in terabytes to exabytes of data—to AWS, you’ll probably lease a Snow Family product. Snow Family products are also good if you need to process data at the edge.
- Anti-tamper and tamper-evident
- End-to-end tracking
- Secure erasure
- AWS Snowcone pricing. You pay to use the device (either an SSD or HDD weighing in at 4.5 pounds) in the form of a service fee, billed by the job. You also have to pay for data transferred out of AWS, but not data transferred in to AWS.
- AWS Snowball pricing. Just as with Snowcone, you pay to use the device and to transfer data out of AWS. You can either pay for this on-demand or at committed upfront pricing for up to 62 percent off the on-demand price.
- AWS Snowmobile pricing. This is a semi truck worth of data—as in an actual semi truck hauling a storage device big enough to take up a 45-foot long shipping container. Pricing is based on how much data you store on the truck each month, and it currently runs at $0.005 per GB per month. But when you’re moving petabytes of data, that quickly adds up. Contact AWS for a quote.
Use AWS Wavelength when you want to deliver low-latency applications across 5G networks. Wavelength actually embeds AWS services in 5G networks, making for a powerful mobile edge computing infrastructure—perfect for resource-heavy tech like VR and network-connected cars.
- Supports t3.medium, t3.xlarge, and r5.2xlarge instances as well as g4dn.2xlarge instances
- Offers Elastic Block Storage (EBS)
- 5G network connectivity
- Resource- and location-based. Being an edge IaaS product, pricing for Wavelength depends on which AWS resources you use and where you use them. Check pricing for the Wavelength Zones you want to operate in, or contact AWS for a quote.
VMware Cloud on AWS
VMware Cloud on AWS is exactly what it sounds like: running VMware’s software-defined data center (SDDC) architecture on the AWS cloud. This lets you run applications in vSphere-based cloud environments while also having access to over 200 AWS services like EC2, S3, RDS, and DynamoDB.
- Disaster recovery
- Integration between the VMware NSX networking platform and AWS Direct Connect
- Compliant with a number of security protocols like ISO 27001, SOC 1 and 2, and HIPAA
- On-demand. VMware Cloud on AWS is pay-for-what-you-use if you choose to pay on-demand by the hour.
- Upfront plans. You can alternatively pay upfront for a discounted rate, with the option to pay for one or three years upfront.
Pricing for AWS compute services
Pay-as-you-go pricing is one of the main selling points of AWS, but there’s a little more to it than meets the eye. For on-demand, compute services are usually billed by the hour or by the second, depending on the product. In the case of Lambda, services are priced by the function and the time it takes to execute your code.
In addition to on-demand pricing, compute services offer a few other pricing options:
- Savings plans—Applies to EC2 and Fargate, offering low prices if you commit to use a certain amount of the services for a term of one or three years.
- Number of requests—For Lambda customers, pricing is based on the number of requests you make and the code execution time. You get up to one million free requests per month.
- Dedicated hosts—A Dedicated Host grants you access to your own physical server. Pricing is on-demand unless you have a Dedicated Host Reservation or Savings Plan.
But no matter the service you use, AWS customers can always expect three things when it comes to pricing:
- Pay-as-you-go. The option to pay only for what you use, usually billed by the hour or by the second.
- Save when you commit. The option to pay a discounted price by committing to use a specific amount of a service for a term of one or three years.
- Pay less by using more. The name really says it all here—these are bulk discounts AWS offers, either for using more of a particular service or for using more services overall.
Some products also offer free trials or free tiers. For example, EC2 is free for up to 750 hours of Linux and Windows t2.micro instances each month for a year, and S3 is free up to 5GB of storage.
Get your head in the clouds
As you well know by now, there’s lots to explore when it comes to AWS compute services. Depending on your prior experience with the cloud and what you’re looking to get out of it, the next step is up to you.
You can take advantage of one of the many free plans or trials AWS offers to get your feet wet, jump right in, or look into pursuing an AWS certification to learn more about the specific services you’ll use in your role. There will be plenty to learn no matter where you start.
The #1 platform for cloud jobseekers
Create your Nerdly profile today, and join a rapidly growing community of the best cloud professionals on the market.
In today's fast-moving digital world, businesses need to keep their apps up-to-date with the latest tech changes. "App Modernization" isn't just a fancy term anymore; it's crucial for companies wanting to stay competitive. NerdRabbit's Nerdly Select Service has...
The global business landscape is rapidly migrating to the cloud as organizations are finding new ways to reduce operational expenses with digital transformation. The cloud offers unparalleled flexibility, cost-efficiency, and boundless opportunities for innovation. As...
We often see discussions of cloud transformation framed as on-premises vs. cloud, but the reality is a bit more nuanced than that. The fact is, you have multiple options when it comes to how your organization will use the cloud, with two of the most popular being...