Looking at the two diagrams shown lower down in the article, I can't be the only one that thinks the "Before" looks infinitely superior to the "After".<p>Yet another third party service to depend on which has unknown reliability or future business model plans.
If you’re familiar with Atlantis and think that this sounds similar - you’re right!<p>They’ve also hired the Atlantis developer: <a href="https://medium.com/runatlantis/joining-hashicorp-200ee9572dc5" rel="nofollow">https://medium.com/runatlantis/joining-hashicorp-200ee9572dc...</a>
It would be great to have an intermediate step between "you are on your own" and "ENTERPRISE – contact us!". Some 'business' plan, perhaps? I know there's 'pro', but it still has the 'request info' button instead of pricing.<p>I am hoping this is what they want to do with the SaaS offering.
As a long-term user (2014/2015), I shall say this is nice but... it is kind of late.<p>I always wondered why Terraform Enterprise / Atlas was in its simple incarnation a tool to be used in CI/CD tooling in form of a containerized app with a selection of storage drivers.<p>That would be a use case applying to hundreds - if not thousands - of companies, and even big players on the market. Those might want the UI with policies too.<p>Who wants to give a tool / backend full access to their infrastructure and not be the one to control all aspects of that tool?
What are the arguments for TF being the better option for a new infrastructure build today?<p>Given that a) Cloudformation with Custom Resources and Macros is extensible, AWS supported, and free and b) its hard to leverage many of the AWS services while still maintaining provider-neutrality (e.g Serverless Framework 2.0 on AWS runs on top of cloudformation templates now)
Terraform is pretty good; but its strength is its weakness; the declarative syntax (DAG planner) isn't perfect and once you start hinting it, it causes all sort of chaos.<p>The solution to complex deploys that re-use portions of the deployments you would want to put a context boundary on, is a series of modules that maintain their own state. But this forces you to pass large amounts of the same data (couples the modules strongly to the main file) over and over again.<p>As for collaborative, terragrunt works pretty good ;) provided you can give a home to the statefile.<p>The real problem I have with terraform/aws (or insert provider) is that its still time-intensive to actually do the deploys. The parallelization of the terraform graph, is "simple" at best.<p>If anyone wants to collaborate on a fork of packer/terraform, let me know on thread.