I tried to calculate a cost estimate for Amazon CodeCatalyst for our comparison page and was befuddled by its complex pricing model. Since I thought I might not be the only one, I summarised my main questions and answers here.
What is Amazon CodeCatalyst?
Amazon calls its CodeCatalyst a “software development service”, which offers a range of cloud and automation resources for devops tasks. One of these services are Dev Environments, i.e. Cloud Development Environments or CDEs. CodeCatalyst is a part of AWS’ cloud platform and only available as SaaS.
What are Amazon CodeCatalyst Dev Environments?
Amazon CodeCatalyst Dev Environments are single containers that are deployed for individual developers, which serve as development environments. The containers typically contain a copy of the source code repository and can be accessed with AWS Cloud9, VS Code or any of the JetBrains IDEs via the browser.
Amazon CodeCatalyst Dev Environments are not intended to build and run the software that developers work on with their Dev Environment. Instead, workflows can be set up where e.g. a commit that a developer does on their Dev Environment triggers a build pipeline in AWS and deploys the software into another AWS resource, for example, a staging environment.
How much does Amazon CodeCatalyst cost?
Like all other AWS products, CodeCatalyst has a fairly complex pricing structure.
Licences are purchased per user. There are 9 different individual aspects for which users are charged. There are currently two licence tiers available, which come with quotas for each of the individual elements in the pricing model. The standard tier is intended for individual developers. In the enterprise tier, quotas can be shared across users.
What are Amazon CodeCatalyst Dev Environment Hours?
When I first tried to do a cost estimate for CodeCatalyst, I was most confused by the fact that the pricing lists separate charges for “Dev Environment hours” and “compute”. The Dev Environment hours is the time that a Dev Environment runs – counting from the time a developer starts a Dev Environment, to the time this Dev Environment is stopped or deleted.
Do Amazon CodeCatalyst Dev Environments cost the same for all machine types?
There are different machine types available for Dev Environments. In the pricing table, a number of hours for Dev Environment hours included in each licence tier is listed, but it does not say if that is independent of the machine type. So I asked. Unfortunately, I got sent in circles through the AWS support system without getting a satisfying answer. In the end, I just assumed that the logic would be the same as for the compute minutes: The hours included in each licence tier are based on the smallest machine type (2 vCPU/4 GB RAM). Each larger machine costs twice as much as the previous one. This is an assumption!
For example, in the enterprise tier, 160 hours of Dev Environment hours are included per user. This would mean:
- 160 Dev Environment hours for a 2 vCPU/4 GB RAM machine, then $0,12/hour
- 80 Dev Environment hours for a 4 vCPU/8 GB RAM machine, then $0,23/hour
- 40 Dev Environment hours for a 8 vCPU/16 GB RAM machine, then $0,46/hour
- 20 Dev Environment hours for a 16 vCPU/32 GB RAM machine, then $0,92/hour
What is Amazon CodeCatalyst compute used for?
Amazon CodeCatalyst compute is used for so-called workflow actions. Workflow actions are automations that can be triggered from within a Dev Environment. For example, when a developer commits code to their source repository in their Dev Environment, this can trigger a build and test workflow. Separate AWS resources are spun up to execute this workflow: a build server, a test environment, etc.
So Amazon CodeCatalyst compute is time that other AWS resources run, in addition to a Dev Environment, which execute automation steps which are triggered by things developers do in their Dev Environment.
What are Amazon CodeCatalyst compute minutes?
But: Amazon CodeCatalyst compute minutes are not actual minutes. There are different machine types available that can be used for workflow actions which use compute minutes. The Amazon CodeCatalyst compute minutes are based on the smallest machine type: 2 vCPU/4 GB RAM. There are three sizes available, and each size costs twice as much as the previous one.
For example, in the enterprise tier, 1500 compute minutes are included per user. This means:
- 1500 minutes compute for a 2 vCPU/4 GB RAM machine (25 hours), then $ 0,005/min ($ 0,3 / hour)
- 750 minutes compute for a 4 vCPU/8 GB RAM machine (12,5 hours), then $ 0,01/min ($ 0,6 / hour)
- 375 minutes compute for a 8 vCPU/16 GB RAM machine (6,25 hours), then $ 0,02/min ($ 1,2 / hour)
What are Amazon CodeCatalyst Custom Blueprints?
One aspect that seemed relevant to me even though it did not have a separate price tag was the “custom blueprints” feature. Researching about it, I was surprised to find out that AWS only recently started supporting custom blueprints, which are a way for organisations to customise their Dev Environments and automated workflows associated with them.
How to calculate a pricing estimate for Amazon CodeCatalyst
To calculate how much CodeCatalyst might cost you, you will need the following information:
- How many hours do you expect your developers to use Dev Environments per month?
Factors that influence this:- How much time your developers spend actually working on code, vs. other activities such as meetings, supporting other colleagues, etc.
- How many different projects your developers work on. If they work on several projects in parallel (i.e. different software products / apps, or different versions of the same software), they might have higher Dev Environment usage because they are likely to run several one Dev Environments at one time.
For our calculation, we assume 160 hours of Dev Environment runtime per month.
- What machine type will your developers need for their Dev Environment?
Factors that influence this:- How large and complex is your codebase? Large and complex code bases might need bigger machines to support even standard IDE functionality such as linting, full-text search, etc.
- What will your developers do within the Dev Environment? If your developers want to compile and debug code on their Dev Environment, this will require more resources than if they only store and edit code.
For our calculation, we assume that we want to run at least part of our application within the Dev Environment and will need an 8 vCPU / 16GB RAM machine.
- What machine type is required for the workflow actions your developers will use, such as build and testing? If you already have build servers or CI/CD pipelines, you can check the machine sizes used there.
Our own build servers run with 4 vCPUs and 16GB RAM, which is not an option available in CodeCatalyst, so we also assumed an 8 vCPU / 16GB RAM machine for compute. - How many workflow actions will your developers need, and how much time do they run? This was the hardest to estimate for us. Our developers build our software locally very often. Our average full-time developer pushes to staging 100-150 times per month. Each build can take anything between 5-30 minutes, depending on how many components need to be built, if the caches are already filled or not, if and which set of tests is run etc. All of this depends on the type of changes that were pushed. I ended up with a minimum boundary of 10 hours of build & test time per month (100 builds, 6 minutes each), with an upper boundary somewhere around 40 hours of build & test time per month (150 builds, 16 minutes each).
For our calculation, I decided to assume 25 hours (= 1500 minutes) of build time per month.
- For our calculation, I decided to assume that the included quotas for storage, package requests etc. will suffice and did not estimate those separately. If your software works with large amounts of data, this might be another relevant cost factor for you.
Our estimate for Amazon CodeCatalyst
| # | Dev Environment hours | Compute minutes |
|---|---|---|
Needed | 160 / moth
8 vCPU / 16GB RAM | 1500 / month
8 vCPU / 16GB RAM |
Included in license | 40 | 375 |
Additional units needed | 120 | 1125 |
Cost per additional unit) | $ 0,46 | $ 0,02 |
Total | 120 * 0,46 = $ 55,2 | 1125 * 0,02 = $ 22,5 |
Grand total:
Basic licence cost: $ 20
Additional Dev Environment hours: $ 55,2
Additional compute minutes: $ 22,5
Total cost per developer per month: $ 97,7
Amazon CodeCatalyst: Who is it for?
The impression I was left with just from researching the pricing model: Amazon CodeCatalyst is a complex product which is targeted primarily at developers and organisations who already use AWS as their main cloud provider, and have already set up their CI/CD pipeline in AWS. For anyone who does not already use AWS for DevOps automations, CodeCatalyst has a steep learning curve.
Comparing their pricing model to others, they are also on the expensive side (though I suspect that many younger CDE vendors currently offer investment-subsidised prices which are not sustainable).
Find out more about the different CDE products currently out there in my whitepaper: Full list of CDE vendors (+feature comparison table)
