Atlas General Availability | Hacker News Reader

Atlas General Availability

hashicorp.com

68 pointsipedrazas36 comments

We use Vagrant for dev and Consul in our deployment and I still have no real idea of what Atlas is.

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?

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.

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.

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

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.

Great feedback. We've heard this and we've been making strides to improve it. For the most part, this is largely our own doing: we didn't do a good job at all of marketing Atlas to the open source community. There are various reasons for this, including not wanting the community to feel we were pushing commercializing onto them. Another reason is that as Atlas was in tech preview, we wanted to build more features and stabilize on the features we had. But we're carefully learning what the boundaries are and are spreading them a bit more.

If you're an open source user, the first thing that you can do is just go to atlas.hashicorp.com and see the features we add on top of the project you're already using. We now break down features by project being augmented.

Example: Vagrant augmentation: https://atlas.hashicorp.com/learn/vagrant

We have pages like that for all our tools.

We'll continue to improve this! Thanks for the feedback.

EDIT: Sorry, I was answering "why don't I know what Atlas is" versus the answer that was expected for "what is Atlas". The best answer for that is to use the homepage and click on the product (Vagrant, Packer, etc.) you use the most! http://atlas.hashicorp.com

I'm interested but still not clear on what this is. And: I'm one of those people that periodically makes a point of reloading your website to see if you're doing something else interesting. I've seen "Atlas" for awhile. Still no idea what it is.

So is Atlas the "Pro" versions of your core offerings?

Or is it something else, a single coherent product that changes the way I'd interact with Vagrant, &c? Is it like Packer As A Service? Does it host images I make with Packer? Is it a centralized Consul repository?

I think it's the "managed nodes" pricing model (which is fine, don't get me wrong) that makes me confused about what the offering is.

Sorry, let me answer that.

Atlas is two things: enhancements to individual open source projects ("pro" version if that helps), and a unification between our projects to give a full dev to prod pipeline.

Per open source project we have, it adds features on top of it we felt didn't fit within the scope of the open source project itself. Example: Vagrant box hosting, Vagrant share (requires a server), Packer builds, Packer artifact hosting, Consul UI, Consul alerts, Consul alert history, Terraform collaboration, Terraform state storage, Terraform run locks, GitHub triggering Terraform, etc. These are all features that enhance an individual HashiCorp project. I'm not going to explain each here, since they're explained in the blog post and of course expect an understanding of that individual project.

Then, Atlas unifies them: code push (Git, CI, etc.) triggers an automatic Packer build which triggers a Terraform plan which triggers a Slack message asking for approval to deploy which can then be deployed with a single button which then causes the Consul UI to update with the latest info.

You can do this with our open source, purely, and many do. This was the inspiration behind Atlas: the companies adopting our open source projects want a complete story, and they're building it on their own, but they'd rather buy it from us. Atlas is us delivering that full dev to prod story.

If you're a HashiCorp user, Atlas is valuable if you're using an indivual project OR if you're looking for a complete deployment solution.

I hope this helps a lot more.

Yup. Neato!

You wrote a fairly long reply, but failed to actually say anything whatsoever about what Atlas is...

mitchellh, you still haven't answered what is Atlas even after your edit. I think is a safe assumption that people have in fact visited your web site and as such are still asking the question what does this do.

If you note, this is the #1 voted comment and you still have not answered the question.

So please take this as an opportunity clarify succinctly what Atlas does and also push your team towards updating your web site.

What makes me sad is that they have been iterating on the site heavily over the past six months, yet they're still dealing with a large amount of confusion about what Atlas even does. I am all in on the Hashicorp stack (we use Vagrant, Packer, Terraform and Consul) and really want this product to succeed, but I am still confused about why I would use it.

edit: as far as I can tell, Atlas is designed for teams to better coordinate infrastructure changes. You can see when someone is building new images in Packer, or deploying changes via Terraform. You can audit these changes, and avoid conflicts

Atlas is just a unifying web UI for all of their other products, and provides a nice overview for your infrastructure.

I think the screenshots on the homepage can explain what it does easier than anything else.

Had the same question, basically read this: https://atlas.hashicorp.com/help/intro/use-cases

For any of us who have watched this team's output and quality and wondered _Why isn't this team being acquired?_, here is Hashicorp's play. Congrats to team on shipping. It will be interesting to see how all this plays out.

I've never used any of Hashicorps stuff (thought consul did arrive on my radar theother day..). Any recommendations? We use the Atlassian stack here (stash, bamboo, etc) and deploy to AWS (regular AMIs rather than any of that fancy docker stuff). Anything here we should look into?

Vagrant has completely changed my development workflow.

So much so that I created a FOSS that uses it and Puppet: https://puphpet.com

That's an awesome project you've got there. Thanks for sharing!

I don't really use anything of Hashicorp's except for Vagrant. That his been fairly revolutionary for all of the teams I've brought it to. That is, Vagrant + Ansible. A repository of Ansible configuration scripts that can easily be slurped down to provision a Vagrant image means much more portable development environments.

In the case of the company I'm currently at it's even more important because we have a firmware build system with a non-trivial configuration that would take days to replicate correctly (because of all the "missing pieces" required that the original developer forgot to write down when documenting the process post-setting it up).

Vagrant. Here's a little brain dump on it @ https://sysadmincasts.com/episodes/42-crash-course-on-vagran...

Packer is great for generating AMIs, particularly if you're already using some form of config mgmt. It will let you tie in your existing puppet/chef/salt/ansible/shell/etc. and output amis based on your config.

Kudos to the team though, love the projects and will continue to support it as it continues to mature.

There's still some pretty significant pain points, bit surprised they released already. Hope they get piles of cash to roll in for all their hard work.

One question I have; with general availability, it seems they indicate packer builds inside of Atlas are free now? It was my understanding that this would be a paid product, so I have my workflow for the OSS vagrant boxes I build all local.

If I can switch my workflow to build and store all the box versions inside of Atlas, that would save some time and effort...

The Packer builds, Vagrant boxes, and Vagrant share are free

$40 per node!

After 10. So more like $40 per 11 nodes, or $3.6

But if I have 15, then it's $13.3 per node. And at my work place we have >100, which means >$36 per node.

That's a lot.

Yes it is.

thought the same!

Pricing seems a bit Atlassian: almost free if you're small, very expensive if you are not.

That sounds more like Github's Enterprise product, which used to be a cliff of a few thousand bucks to fall off, where private Bitbucket is a $1/user/month. Maybe Github's pricing is less unusual now.

Github is still incredibly expensive, for both Enterprise and Github.com.

While $40 per node might sound high, I think it is reasonable. I've worked on several products that are priced per node/server.

Per-node a difficult proposition -- the price of "per server" of something varies wildly, for example a t1.micro is roughly $9.50 a month, while a m4.4xlarge is $738. $40 on 9.50 seems extreme, but $40 on $738, and Atlas is only a 5% of the cost of your server.

But when your target customer is using high end instances, because they are creating more value for their company that way, a 5% upcharge for better management, can be justified. Mostly its because the team of ops/software engineers supporting production environment are going to cost more than the infrastructure anyways.

Additionally, any large deal (>1000 nodes) is gonna get negotiated anyways, so public pricing is kinda a wash.

I appreciate the clear, up front pricing. How does the pricing compare to the competition?

Huzzah! Excited to use Atlas on my current project. We're supposed to come onsite to talk with Kevin next week. Looking forward!

Congrats Mitchell + Anand + Kevin + team!

Now send someone to softwarecircus.io to talk about it!

Reply on news.ycombinator.com