Both CloudFormation and Elastic Beanstalk are a way of deploying your application on AWS.
But they’re fundamentally different approaches. Beanstalk is PaaS (platform as a service) while CloudFormation is IaC (infrastructure as code).
Beanstalk is similar to tools like Heroku and Engine Yard – a way of making provisioning easy.
CloudFormation is the opposite! But once you manage the complexity, it has a lot to offer.
In this piece, we’ll be exploring the differences and similarities, and helping you work out which is the right fit for you.
IaC vs PaaS
If you want to get straight to the toss up between Beanstalk and CloudFormation, you can skip this part.
But if you’ve never heard these terms before, or pronounce IaC like I-ack , then it’s time for a history lesson.
IaaS to PaaS
aaS, stands for as a service, meaning a service delivered via the web. The first part of computing to be offered this way was infrastructure (IaaS) which is the very basis of all cloud computing.
IaaC services like EC2 or Azure VM were virtual versions of the bare metal servers all applications used to run on.
The next step was to offer the layer above the server – things like OS, database management and dev tools – as a service, too. This layer is called the platform, being neither the application or the infrastructure.
Beanstalk fits here, and its job is managing the infrastructure beneath it.
So what’s IaC?
IaC is not quite the same thing as IaaS, although it does relate to infrastructure management.
IaC refers specifically to the capability to specify infrastructure as code, the IaC service then provisioning the infrastructure described in that file.
So something of a what-it-says-on-the-tin service.
Now we’re clear Beanstalk and CloudFormation are ultimately both ways of managing infrastructure, we can drill down to the differences.
Beanstalk vs CloudFormation In a nutshell
The essential difference between these services is that Beanstalk handles deployment/provisioning for you while CloudFormation requires a lot of input.
So one’s low effort, low control, and vice versa.
image credit: stackoverflow
How Beanstalk works
Beanstalk can be accessed via the AWS management console, line command, or API.
Language-wise it supports Go, Java, Node.js, PHP, Python, Ruby and Docker Containers and more.
It can be set-up without very much input at all, although users can specify EC2 instance types, and add additional resources such as RDS.
Once your application has been uploaded, Beanstalk will handle the following with sensible defaults:
- Load balancing
- And application health monitoring
Elastic Beanstalk allows you to fine-tune the details of any provisioned resources if you do want a bit more control.
And you can also add resources such as an SQS queue, CloudWatch alarms, and even an ElastiCache cluster, which can be configured through .ebextensions configuration files.
Plus, you can roll back to previous versions.
- Easy access to automation – the major draw of Beanstalk for most users: it runs your infrastructure with very little input required
- Cost – there’s no additional cost for Beanstalk itself, you just pay for the resources it provisions.
- Flexibility – you do have the option of managing any part of your Beanstalk stack manually, which isn’t the case for CloudFormation
- Not easy to diagnose problems – because it managed everything for you, finding out exactly where Beanstalk went wrong – in the case that it does – can require a lot of detective work
- Updates to stack – Beanstalk will makes changes to your stack inline with updates, and though in many cases this is totally benign, in some, it can cause issues
- Lack of fine control – although many of the default settings on a Beanstalk-deployed app are sensible, they won’t suit every use case
- Flexibility has limits – while you can deploy additional resources using .ebextensions, it tends to feel like you’re stretching Elastic Beanstalk beyond its intended use case.
How it works – CloudFormation
The user writes a YAML or Json template specifying the AWS services used.
Once uploaded, CloudFormation can then create and deploy the specified resources.
AWS resources collected together to run one application are known as a stack, and it’s possible to reuse your template to deploy that stack in multiple environments.
- Brings code review features to infra – because it’s IaC, you can now review your infra changes as you would code changes, allowing for greater oversight (often used in DevOps)
- Granular control – the main benefit of CloudFormation is offering the experienced user granular control and oversight of their AWS stack
- Large community support – although it’s complex, CloudFormation is an established tool tool with a big community and resource base
- Front-loading on learning compared to Beanstalk – a ‘crash course’ in CloudFormation will leave you with gaping holes in your knowledge, which is all the more problematic because…
- User error can pose real risks – CloudFormation makes it easy for a small mistakes to have big impacts, and it’s common for users to accidentally delete entire services from their stacks
- CloudFormation drift – when working with CloudFormation, it is important that all changes are done through CloudFormation. Manual changes lead to drift, and when there is too much drift, the CloudFormation code will be unusable for future changes – meaning the benefits of IaC will be lost.
Now we have an idea of how Beanstalk and CloudFormation work, we can see how each fits different use cases.
Beanstalk is good for
Beanstalk can be great for teams who – for whatever reason – don’t have the time or resources to invest in cloud.
It provides many of the benefits of a knowledgeable MSP – auto-scaling, etc – for free, and with minimal input on the part of the user.
It has some flexibility too, but it works excellently for teams with common application types.
So, it tends to suit:
- Small application-focussed teams
- Teams with a simple application like an e-commerce site
- Start-ups who want to get an MVP out quickly
CloudFormation is good for
CloudFormation is a powerful tool if you know how to use it and you need granular control over your AWS stack.
It’s not the easiest to use, but it opens up a lot of avenues for those who’ve mastered it.
It tends to suit:
- Teams with a lot of workloads who need a way of streamlining provisioning
- Teams with complex applications who need more control over their stacks
- Teams who need to manage their resources more closely for cost or security reasons
- Teams that have dedicated DevOps engineers (or a good DevOps as a service partner) to be the managers and gatekeepers of CloudFormation
How we can help
As an Advanced Consulting Partner, we have the skills and knowledge to take on nearly any AWS project.
We’ve worked with multiple clients using Elastic Beanstalk and many different IaC approaches. We’re also AWS Well-Architected Review Partners, DCX Partners and experts at supporting AWS around the clock, so whatever the workload, we’ve got you covered.
Whether you’re starting out for the first time, migrating, or optimising and supporting an existing infrastructure, just get in touch.
To learn more about AWS services, check out our piece on Redshift vs. Athena vs. EMR.