Indeed - but the Terraform community appears to be hit the hardest. Because unlike Consul, there's no open-source hostable anything. It feels just an arbitrary restriction for a language + CLI; the terms of the licence can be viewed as if whatever you ship has Terraform cli embedded, you're in breach. I can see the reason for SPL like Mongo did - it is indeed unfair for AWS to make money off hosted open-source mongodb. But this move by Hashi I'm struggling to wrap my head around

> it is indeed unfair for AWS to make money off hosted open-source mongodb

Why? If software authors don't understand that releasing their code under an open source license means someone else other than them may potentially benefit from it, that's really on them.

Authors do understand the benefits of OSS development.

Predatory unethical open source freeloaders don’t.

Disclosure: I work for Amazon.

Amazon has never offered a database service based on the MongoDB server software. Not when it was AGPLv3 nor when it was SSPLv1.

But Amazon did copy the software and continues to sow confusion in the community about which is which

No, it didn't copy the software. That's like saying min.io copied S3 software, which would also be false.

It implemented a wire protocol for interoperability purposes, similar to functionality in IBM DB2, Azure CosmosDB, FerretDB, and other databases.

Right and it continues to sow confusion in the community

If there is anything specific you believe is confusing, please provide some details and I will pass the feedback on.

Sure just do a Google search and you will see that a user searching for the database is going to get routed to the clone in the AWS organic and sponsored results

Based on BSL, if you aren't trying to commercialize on top of Terraform you'd still be fine.

If you're using Terraform to manage deployments in test and development it's still business as usual, but if you're actively using them in production or making a company that sells a wrapper around Terraform then you'd have to pay Hashicorp.

Honestly it's pretty fair. I can continue to make my own personal, test, or development apps using vault, consul, and terraform.

And this move was done because plenty of large companies (not naming names because this is starting to touch a legal gray area) are using Vault and Consul in production to generate hundreds of millions to billions a year in revenue.

with Vault and Consul that makes a ton of sense; they're centralised backend services that are inherently SaaS-like. But expanding the "on top of terraform" definition to a CLI / language? That's insane. It's like making a programming language or framework proprietary and saying you can't sell anything written in the language without the language creator's permission. I guess such languages did exist in the early days of the industry but those days are long gone

As has been pointed out elsewhere in the comments, this is incorrect. The switch to BSL does not mean you have to start paying Hashicorp to use their formerly OSS products in production. To quote their FAQ:

> All non-production uses are permitted. All production uses are allowed other than hosting or embedding the software in an offering competitive with HashiCorp products or services.

Doesn't this eliminate any managed runners since they'd compete with Terraform Cloud?

> unfair for AWS to make money off hosted open-source

Competition is healthy. Everyone should be reluctant to lock himself in with a vendor seeking a monopoly.

Yes, people underestimate how important consul and vault are in modern infra. There is no direct replacement for vault or consul in how versatile and powerful these tools are.

I definitely underestimate them because in my last 7 roles I haven't used either. I don't know how far back 'modern' goes, but 3 roles were over the last 24 months (just started a new role) and none of those roles the companies used consul or vault to manage anything.

If you've ever done a credit card transaction or any form of dollar based transaction, it's using software directly built on top of Consul.

Are there public details about this? Are you referring to the Visa Network or something related?

I'm assuming you're missing some words here, probably "with Stripe"

The idea that every credit card transaction or any form of dollar based transaction works on software directly built on Consul is patently false.

“With Visa or Mastercard” are likely the missing words.

That is not the case unless you're talking about the underlying raft protocol.

Ask yourself how much of the industry your experience represents. Should answer your question.

This is why they're having to make these moves. How much value have they given you vs how much have you paid? If the gap is too large, they have to start locking some things down.

Exactly this.

Open Source software quickly becomes a tragedy of the commons because most organizations are unscrupulous. Look at how AWS destroyed the entire Elastic, MongoDB, PostgreSQL, Kafka, etc ecosystems by abusing overly permissive licenses.

Since when is abiding by the license abuse?

Oh you mean the unwritten, unenforceable, undefined, romanticized spirit of the license?

You ain't wrong!

It's a major reason why I'd never release a core revenue generating feature of mine under a permissive Open Source License.

Unsrupulous implies some form of an ethical code, whereby you are judged by intent, not by action - which is an idealized view of the world but an unrealistic one at that.

They're not unscrupulous. They're using the software as it is licenced, which is the software's declaration of how it is allowed to be used.

The problem lies with OSS, not with people using it.

The problem lies with people expecting to apply non-FOSS monetization strategies to FOSS.

Particularly with corporate charity (aka "permissive") licenses. If you want the source to stay open for users further down the chain, use a license that requires that.

One should wonder why companies love "open source" software so much and despise "free software" even more.

On the matter, Grafana has relicensed most of its software to AGPLv3 (eg: https://github.com/grafana/grafana/blob/main/LICENSE)

Strongly disagree with your point as you are comparing SaaS vendors who package these technologies under the hood with no immediate access provided to their customers, just as means to delivering a service with tech providers who were just blatantly packed by AWS and provided as identical technology.

There is a huuuuge difference between OP and your examples.

the tragedy of the commons is a thoroughly debunked though experiment, and actual studies show that in the real world, just like in this case, the real tragedy of the commons is when one entity (hashicorp) is able to monopolize a common good (open source software with contributions from tons of people outside of hashicorp) in a way everyone who was being a good user of the commons (by contributing back to those projects) is opposed to because it actively destroys something everyone was using collaboratively.

I have been thinking a lot about the tragedy of the commons (totc) within the context of usufructian socialist services and expanding them beyond their fairly narrow use case currently (public libraries are the biggest example). I was theorycrafting community makerspaces, tool libraries, and public baths, but I was hung up on thinking about totc. Do you have some sources that you would recommend for more detail on refuting or debunking totc? Thanks!

How has Amazon destroyed these ecosystems? As far as I know, all of these products are very much alive and well.

> Look at how AWS destroyed the entire ... PostgreSQL, Kafka, etc ecosystems

Err, I think you need better examples, as the PostgreSQL ecosystem is doing just fine, and RDS doesn't threaten anyone's IPO price. Since Kafka is not only Apache licensed but is an Apache project, I gravely doubt that the Apache Foundation is going to IPO anytime soon either

If you want to pick on two other ecosystem casualties by the Big Bad AWS Gonna Steal Our Shit fearmongering, use Sentry and Sourcegraph as examples of "we want the community until we don't" rug pulls

Do you think Google should lock down the go language and compiler, considering how much value they've given Terraform and likely how little Terraform has paid for go itself?

Alphabet doesn't generate revenue from Go.

Following your example, Alphabet should open source all the weights they use in their current search algorithm as well as the algorithm itself.

And extending that principle, they should also open source the machine translation model and weights Robert DeNero made for them when at Google Translate.

Both of these are dumb hypotheticals but serve to point out the fact that there are loss leaders and revenue generators at each company.

No. My point was, companies should make money from services, not software. I believe all software should have a FOSS license.

I get your point but wouldn't that encourage businesses to make their software to be as badly designed as possible to sell as much service/support as possible on e they get ahold of the market?

> Alphabet doesn't generate revenue from Go.

Hashicorp doesn't generate revenue from Terraform.

(They do generate revenue from Terraform Cloud, but that's a backend for Terraform, not Terraform itself)

True, just like Qualcomm bought Eudora (mac) email for fame, Google started the Go project to keep Rob Pike (of UNIX fame) happy and to help their own internal SRE teams. The public "good will" and benefits to Google SRE have more than paid for their efforts to create the Go language, plus also there is some publicity and fame for Google having most of the remaining early UNIX guys on staff ...

My employer pays for Visual Studio, JetBrains, Mac tax + dev license pays for XCode, Oracle, SQL Server,...

I guess it is up to Google to decide how to monetize Go, or who knows, maybe some day they will put it into a foundation of sorts that should find a way to keep sponsoring its development.

Terraform would have avoided years of stringly typed mess had it not been written in Go. I’d estimate that half the provider bugs I saw were either because of a lack of generics (and correct unwillingness to generate code), or because of casting empty interfaces.

The only advantage was static binaries, which not many other languages in a usable state supported at the time. It’s not even like there was an ecosystem at the time - HC had to maintain an AWS client for Go for several years, and similarly for Azure.

Do you resell services that would be competing with Vault and Consul? If not I don’t see how this applies to your company’s product.

Side note: I’m not defending the license change and generally think most of HashiCorp’s products are not that great.

You can't use BSL licensed products in production without paying the provider.

Plenty of companies are using Vault and Consul in this manner.

That is not accurate. BSL does not prevent simply using the products, it prevents packaging them into a competing offering.

Review the FAQ here: https://www.hashicorp.com/license-faq

> BSL does not prevent simply using the products, it prevents packaging them into a competing offering.

The problem is that "competing offering" is poorly defined.

Don’t think it’s targeted at Vault. Vault has APIs so in the enterprise, you can self host and a vendor can manage vault through the APIs. Terraform specifically is built CLI only in the OSS so you have to repackage the underlying binary and hence are subject to the license as a vendor.

that's the most puzzling bit. Do they not want to anyone offering anything that could compete with any of Hashi's products to be able to use terraform binary at all?

if so, that's similar to proprietary programming languages. Not a thing. The community can just agree on a similar but open alternative and the original company is left behind. That's why all languages and frameworks are open-source with permissive licenses.

and if not, if it's just about the hosted / managed parts - then what exactly is it that I can use wrongly? Terraform Cloud / Enterprise was never open source. There's nothing I can self-host and charge users for...

> Do they not want to anyone offering anything that could compete with any of Hashi's products to be able to use terraform binary at all?

I think they want to make it difficult for SaaS products that embed terraform (or an altered version of terraform I would assume where they have built APIs) in their product. That said, customers can install terraform so you can still hypothetically have an "integration" with Terraform where you call the raw binary (like Atlantis). So basically, they want terraform cloud to have a "privileged" user experience is my guess.

On the other hand, we wanted to give them money to have a support contract, maybe some yearly architecture reviews or a call with a consultant to point us towards new features. We have similar agreements with e.g. AWS, we're looking at setting something like this up with a postgres consultancy. However, the smallest offering was solidly in the six figures a year. That put it into a very different investment process, the board got involved and you can all guess where it went from there.

I had the same experience, wanted to give them money but not six figures for each service we used.

I'm also surprised no one is really mentioning Nomad in this discussion either, I know for a fact circleci used nomad at some point to schedule CI tasks on worker machines. If I create a SaaS product that runs programs which get scheduled with nomad under the hood is that a breach of the license?

> I know most HN commentators are thinking about Terraform, but I think this change was done with Consul and Vault in mind.

Clearly Terraform was also in mind, since Terraform is now under the BUSL: https://github.com/hashicorp/terraform/blob/main/LICENSE