The Beginning of the End for Terraform? | Hacker News Reader

The Beginning of the End for Terraform?

medium.com

30 pointslockedinspace21 comments

"Who will buy HashiCorp" was the question that everyone was asking for basically their entire existence.

Vagrant, Terraform, Vault, Consul and Nomad were just too good. HashiCorp's biggest competitor against Terraform and Vault Enterprise was their own stuff! If your platform team was even moderately competent, you had no need for any of that stuff, especially at the prices they were asking for!

Personally, I was hoping for Microsoft to acquire them.

Terraform as a replacement for ARM is _literally_ the dream, and Vault would have been a really good underlying service for Azure Key Vault.

Now that I think about it, I'd love a universe in which GitHub is Azure DevOps, TFE-but-better is Azure RM, and Vault is Azure Key Vault. Alas, that is not reality.

I want to think that Microsoft made a bid but lost out to IBM.

All that said, OpenTofu is a drop-in replacement for Terraform; nearly zero work required. I haven't tried OpenBao yet, but I presume that it's the same situation there. There's enough steam propelling them to make me think that they'll pull through for the long haul.

> OpenBao yet, but I presume that it's the same situation there

no, they removed a lot of things, espeically on the storage side.

Bicep has the potential to provide a cross platform solution, but it’s unclear whether that’s the direction: https://github.com/radius-project/bicep-types-aws

I would not expect that to happen any time soon. Maybe they will see a chance to promote bicep more by such move. Yet until now they actively (and reasonably) advise in their docs Terraform use, if you consider multicloud and few other situations.

I would rather expect MS to declare support for OpenTofu. They don't even have to be involved in it directly.

The main work is ensuring azure providers compatibility (which is formally maintained by hashi). Add compatibility with Tofu to Azure Verified Modules [0] (a great project btw), and we have a clean way out of this by providing real help where it's most needed.

[0] https://azure.github.io/Azure-Verified-Modules/

I can reassure on one point - why would IBM support an inherently multi-cloud platform?

Because at less than 2% market share they are a beneficiary.

The main reason orgs move to multi-cloud is they have to support another cloud for business reasons - AI, cost optimization, etc.

When they replatform internal ops to Terraform any cloud provider can become a viable option. Beforehand, not at all.

Why IBM might not be successful with this acquisition, it is likely the only way they could really grow market share.

If anything Terraform will get reinvigorated. I'd bet IBM will reverse the license change by the end of the year.

> This leads me to my first concern, IBM has a conflict of interest. IBM has a cloud offering, admittedly with a 1.8% market share. Why would they want to keep developing tools that frankly benefit their competitors more than them?

Because they'll make bank off of Hashicorp Cloud Platform.

> Another option is to ‘go native,’ with each of the big three hyperscalers offering its dedicated options: Cloud Formation with AWS, Bicep with Azure, and Config Controller in Google Cloud.

Not an option. People manage way, way more than just AWS, Azure and GCP with Terraform. GitHub, Cloudflare, Tailscale, etc ...

It's an interesting time for Hashistack for sure, but I don't think the industry is prepared to move away from the tooling.

> Because they'll make bank off of Hashicorp Cloud Platform

I’m curious what you think IBM will do differently to make bank, considering Hashicorp failed to make bank from their cloud platform offerings.

They’ll sell it

I guess most companies now a days just say open-source to get some initial traction, nut no worries it will get a fork like redis did if they change the license.

There's already a fork, OpenTofu https://opentofu.org/

Why is it happening so often now? Most OS companies changing their licenses...?

Because historically they made money selling managed version of their services.

Now GCP, AWS and Azure just sell a superior managed version for cheaper. So they lost their income stream.

Open source products tried to change the license so the big guys would at least have to pay them something to make money on their work.

Isn’t that problem easily solved by adding to your open source project’s license: “FAANG companies (like Google, etc.) cannot use or make profit from this project” ?

Better stated, of course, but you get the point.

That change makes it no longer free/open source.

They can just open subsidiaries and use that to go around them. Also the big guys do donate a lot of money.

The real reason might be VCs. The companies going closed source have raised from VCs and they are pushed to more and more growth just changing licenses and taking control.

There isn't a well known / widely accepted license with such clauses (afaik, happy to be corrected). Noone other than individual tinkerers will ever touch anything with a custom license or any license ambiguity for that matter.

Meta's LlaMa for example is "OSS" but under a custom license which effectively prohibits hyperscalers from using it

I mean that's basically what they're doing. The BUSL license these projects are using boil down to "you can't resell the project against us". But that's a non-starter for many folks and the companies using such a license are the devil.

Yes, there is far too much ambiguity in these licenses. Especially because you never know what will happen in the future. You could be found to be in competition after they launch a new product or new feature, years after not being in competition. At that point changing out their product could be too challenging, forcing you into arbitrarily expensive commercial license agreements.

Certainly one of the reasons why open source is popular in software is that it gives you options for maintenance prices. You can do it yourself or pay any number of consulting agencies to do it for you, or the creators of the software. When it becomes locked to one vendor suddenly the market economics change hugely. Now that vendor can crank the price up extremely high, basically until just before the point your willing to engage in a hugely expensive software refactor to move to a different product.

Open standards were supposed to help make it easier to move to new products, but in reality, it rarely is that clean. E.g. look at SQL, while it's easier maybe to move from one SQL database to another than from one completely custom database to another, it's still a massive amount of work due to details of each SQL server.

That's exactly what Terraform and Redis did.

The open source license is designed to guarantee your liberties. The moment you start to remove some liberties you have neutered it.

such software cannot be included in many popular linux distributions which means less people are going to have easy access to it.

i found this useful for related general background information as well as some interesting conclusions, though you should probably decide for yourself if you think they are something you agree with.

https://www.jeffgeerling.com/blog/2024/corporate-open-source...

If you prefer video, this is actually a transcription of the youtube video here :

https://www.youtube.com/watch?v=hNcBk6cwim8

Reply on news.ycombinator.com