django-step-by-step VS jsii

Here's an example of a docker-compose file in my reference Django project: https://github.com/briancaffey/django-step-by-step/blob/main/docker-compose.yml

Mono repo with a sample Django micro blogging app (μblog) and frontend app (Vue SPA written with Quasar), GitHub Action workflows for infrastructure and (separate) application deployment pipelines, IaC code that consumes each of the libraries listed above, VuePress documentation site and miscellaneous items (k6 load testing scripts, Cypress tests, docker-compose, etc.): github.com/briancaffey/django-step-by-step

This project also uses a model manager to make it easy to show how many people like a post as well as if the current user likes the post or not https://github.com/briancaffey/django-step-by-step/blob/main/backend/apps/blog/managers.py.

Here's another repo that consumes these IaC libraries and deploys them through GitHub Actions pipelines: https://github.com/briancaffey/django-step-by-step

By ignoring changes here you don’t have to worry about infra upgrades pushing out application updates. And you can update your app with with a separate pipeline. My code uses AWS CLI to register new task definitions. Here’s an example of my pipeline that deploy the app: https://github.com/briancaffey/django-step-by-step/blob/main/.github/workflows/ad_hoc_update_backend.yml

Here are some other questions that would be good to ask about the infrastructure: * What AWS account will this run in? Who has access to that account? Are SSO permission sets used? * What DNS records will be used to access the frontend and the backend? Are these two services accessed via the same record URL? For example, all do you to only send example.com/api/* traffic to the backend and then send all other example.com/* to the Angular frontend? Or do you want to use api.example.com for all API traffic and example.com for the Angular site. What about non-prod environments? Will you use alpha.example.com for a staging environment for a non-prod environment named alpha? * You mentioned that the team will be using Terraform heavily, how will you be splitting up your modules? When you run Terraform's apply command, does it update a single set of infrastructure for your entire application, or do you run terraform apply several times for different groups of resources, such as a networking stack (with VPC), data stack (with RDS) and application stack (for ECS resources), for example? You will probably use terraform remote state to manage these different logical components and the data dependencies they have between each other. * Another related question is how you are running pipelines for Terraform? I would use something like GitHub Actions. When a pipeline runs, you should see the output of a terraform plan stage and then the pipeline should pause for manual approval after the changes have been reviewed. This can be done with GitHub environments, for example, and other CI/CD tools can also do this. * What happens when you need to change an environment variable? This can be a complicated question. Environments that are not secrets might be stored in a terraform.tfvars file, or might be stored in environment variables in your pipeline in the form of TFVAR{name}, so this implies that changing an environment variable is an infrastructure change. You want to keep your infrastructure deployments separate from your application deployments. If you are using ECS, you probably want to use ignore_changes for the task definitions referenced by the services. This will create a new task definition, but it will not be used by the new services. When you do an application update, the service may use this new task definition with the updated environment variable. * Another question: does anyone on the team need direct access to the database? Is there a bastion host used in the infrastructure that can people can connect to the DB with via port forwarding? I hope these questions are helpful. I have thought about a lot of these and have been exploring their answers with an open source project focused on a containerized, database-backed Python web app (made with Django) that I deploy on AWS with ECS Fargate, and I have another repo with Terraform modules for deploying application infrastructure. Here's the application repo: https://github.com/briancaffey/django-step-by-step and here is the Terraform module repo: https://github.com/briancaffey/terraform-aws-django. Please let me know if you have any questions

Here is how I organize my modules for an example microblog app: https://github.com/briancaffey/django-step-by-step/tree/main/quasar-app/src/modules

Yes, this is possible. There are different ways to do this and it will depend on how and where you host your applications. Here's an example project of mine that I run on AWS that uses some the same URL with some routes for Django template-powered views, some views for a Vue app, some views for REST API and some views for GraphQL: https://github.com/briancaffey/django-step-by-step.

Here's an example of how I implement CI and CD with a Django project on AWS using GitHub Actions. I use ECS Fargate. https://github.com/briancaffey/django-step-by-step/blob/main/.github/workflows/backend_linting_and_unit_tests.yml

What about jsii? The technology behind AWS sdks: https://aws.github.io/jsii/

Is Stainless similar, different?

The libraries for other supported languages of CDK are built as wrappers for underlying JS or maybe TS code via https://github.com/aws/jsii

So all the core CDK code is written first in JS/TS and then stubs for the other languages are added

Unfortunately this is often done without consideration for how the other supported langs actually work, and artefacts of e.g. JS lack of support for kwargs leak through

This is why e.g. the typing in CDK Python is completely broken - pretty much uniformly the concrete types like "Resource" don't implement their corresponding interface like "IResource" (to a type checker)

(There are many other typing niggles like this but that's the most egregious and pervasive one)

At the end of the day, having to explicitly cast concrete types as their interface to satisfy type checker is a minor annoyance, albeit a stupid one that could have been avoided with more care in the core library.

I could live with that, but I encountered so many bugs and issues trying to use CDK on current project that it's now much clearer to me why every company I worked at previously was using Terraform.

Pretty sure some of those issues are ultimately CloudFormation ones. The cumbersome CF > CDK JS > CDK Python stack is great for obfuscating errors and making debugging hard or impossible though.

Pulumi do something similar, albeit with Go as the core language and Terraform underneath. From what I've seen with a little use they have a much more successful result though, Pulumi Python was not a complete mess, and deploys faster and more reliable with better error feedback. I guess they just took more care to get it right.

Every version of the CDK uses an interop layer and runs on top of the Typescript version

https://github.com/aws/jsii

And as far as TF supports services before CFT. Guess which is easier for an AWS employee to do - getting the CF service team to support a new service or just contribute to Terraform’s open source project?

I know of at least one service where the service team introduced the needed APIs and then an employee of AWS wrote the TF provider and contributed to the project before AWS’s own internal team added it to CFT.

Source: former AWS ProServe employee

To "meet developers where they are" is a beautiful tenet of AWS, and of the CDK, and inspired us to create awesome technology such as JSII and constructs.

Yeah both CDKTF and Pulumi piggyback on JSII, a class translator that supports the languages you've mentioned, which was used as a building base for CDK.

The reason we chose to start with compilation to JavaScript and not some other popular cloud language is because JavaScript is currently one of the most used ones, and in addition, we are using JSII to write our SDK in Typescript and then export it for consumption in other languages, such as Go, Python, Java and others.

The CDKTF CLI is implemented in TypeScript. Providers and resources are implemented in TypeScript as well. jsii is used to compile the providers and resources to the supported languages.