[seiji]

Good rationale for not having the people who write the product promote the product. You're too close to remember everybody else doesn't know what things are.

Based only on the leading description, it sounds like something trying to be a non-crap Docker replacement that was actually built from experience instead of just trying to capture a VC money waterfall? Atlas builds upon and unites our popular open source tooling to create a version control system for infrastructure. Automate, audit, and collaborate on infrastructure changes across any infrastructure provider.

Though, looking more at this post, it's a "we did this product" post, not a post about the product: the product has become a polished, intuitive experience.

That's nice?

The actual product page starts with: Atlas unites HashiCorp development and infrastructure management tools to create a version control system for infrastructure. Automate, audit, and collaborate on infrastructure changes.

Which, I'm still not sure.

Atlas can be used for all its features, or features can be used individually. All components are HashiCorp open source tools that are hosted and run in Atlas.

That's what it can do, but not what it do. Is this a side effect of SF startup culture? Do we forget everybody doesn't live inside our startups?

As best I can tell, Atlas is their map (heh) of how all their pre-existing products work together, and you can buy one umbrella product/support instead of buying 4 individual products?

[Gigablah]

I'll take a stab at this.

Atlas is an application that glues together four different projects by Hashicorp: Packer, Vagrant, Terraform and Consul.

Packer creates virtual machine images (also has Docker support). Vagrant will create, provision and run them for you (so a bit of overlap with Packer there). Terraform is basically a platform agnostic version of Amazon CloudFormation (create and manage infrastructure via configuration). Consul is a service discovery solution.

All of these tools used in conjunction can help you set up a machine, or a cluster of machines, on any supported provider, that will run your application images, all from a set of configuration files. More importantly, changes to your infrastructure and application environment are now a matter of editing configuration.

Atlas places a UI in front of these tools and allows you to execute commands and monitor their status.

[tptacek]

So like, this is Hashicorp's answer to Kubernetes.

It's a sweeping ambitious suite of tools that I'd use to essentially create my own Heroku out of raw machine instances, of my own or from commodity providers.

[Gigablah]

And Atlas is to those suite of tools like how GitHub is to git, kinda.

[seiji]

That makes more sense, especially the one line description: Atlas places a UI in front of these tools and allows you to execute commands and monitor their status.

Companies are willing to pay for better user interfaces and better monitoring.

But, have we forgotten about hello world?

What this really needs is a barebones absolute minimum-step demo page showing "how to reliably deploy todo list webapp" for immediate benefit. Then maybe add failover and HA and other more useful minimum-step demos building on top of prior examples. (Maybe it exists? One didn't jump out at me and I didn't click around much to discover further.)

Show, don't tell.