Earlier quoted context omitted.
None of this is new. GNU was founded and developed its licenses to address this issue. But there’s a tradeoff. One that GNU happily accepts. But there has never been a “code of honor” in this space. Give me a break. Companies have been offering Apache- and GPL- licensed software as a service for longer than Elastic has been a company. Elastic wanted to have its cake and eat it too. They overplayed their hand here and…
People getting mad about a code of honor in a decades old, brutal business environment is truly bizarre to me. This isn't a playground pickup game, it's business.
AWS announces forks of Elasticsearch and Kibana
621–630 of 969 posts
Re: AWS announces forks of Elasticsearch and Kibana
#622The issue here is that it's quite impossible to compete with AWS if it offers even an inferior version of a service. Because of network egress pricing. You are most likely on AWS already. And even if elastic offers a 2x better product at 1/2 the price, your cost of traffic for an external service will easily 4x versus using AWS services. And that's really the frustration for these service providers. You can NEVER com…
Re: AWS announces forks of Elasticsearch and Kibana
#623Earlier quoted context omitted.
I agree with the other comments here saying that this is not a gap in open-source licensing. If Elastic wanted to force Amazon to contribute modifications back, they could have switched to the AGPL to do that. But they didn't — because, as the AWS blog points out, AWS was already contributing their changes back. The problem wasn't that the spirit of open-source was violated. The problem was that AWS is better at acqu…
Maybe there is a middle ground for companies to use SSPL/BSL with a "profit sharing" clause or similar. It's still not "open source" in the traditional sense. But it could be a "win/win" scenario?
My main problem with the pro-Elastic arguments it that the crux of the argument is:
1. Elastic drives the vast majority of development for ES 2. Elastic needs revenue in order to keep paying devs to work on ES 3. AWS is competing too well with Elastic's hosted ES
And there is an underlying implication that if Elastic fails, ES development will stall almost completely and a great software project will die. But like... isn't that the point of OSS? If a software project lives and dies with a single company, what exactly is the point of it being OSS?
Re: AWS announces forks of Elasticsearch and Kibana
#624Earlier quoted context omitted.
> AWS knows that if they take too much of ES' market that they won't survive. If they don't survive it will just be a matter of time before ES is dropped by Amazon and totally unsupported. Or Amazon (or a third party) may take over maintenance of ES if it's worth it to them. That's the whole point of open source. > You can frame this question in terms of ethics, you can frame it in terms of licensing naivety, you can…
> Amazon is outcompeting them by providing a better service to their clients. I wouldn't necessarily say AWS' ability to use ES as a loss-leader for their Data Transfer charges is better for AWS's clients in the long-run. I can certainly see the argument for calling it exploitation.
Re: AWS announces forks of Elasticsearch and Kibana
#625Re: AWS announces forks of Elasticsearch and Kibana
#626Earlier quoted context omitted.
The AGPL is nearly 2 decades old, and its tradeoffs are well known among people working in open source. If Elastic had wanted its protections (and restrictions) they easily could have chosen it.
AGPL also harms adoption at companies because of the "invasive" nature of the license. Google, for example, bans the use of any AGPL software. It's too vague for the legal team to confidently understand the "bounds" of the copyleft license. Permissive licenses like MIT/Apache have historically been what companies prefer to have when evaluating free software. And, generally, it hasn't been a problem for companies like…
Re: AWS announces forks of Elasticsearch and Kibana
#627Earlier quoted context omitted.
Is the AGPL not a copyleft license? What are the derivative works you're required to distribute as AGPL?
If you're using a database that's not tightly integrated into your project and distributed with it, I don't believe you'd have to open your project. The API is the boundary: - your project running on Linux kernel doesn't become GPL - your project talking to MySQL doesn't become GPL and one talking to mongodb pre-2018 doesn't become AGPL - your project talking to elasticsearch wouldn't become AGPL (However AWS search…
Re: AWS announces forks of Elasticsearch and Kibana
#628Earlier quoted context omitted.
Popular OSS licenses were designed in an era to allow direct customers (businesses) to directly install and use the software, and redistribution means only distributing the package. Nobody at that time knew or predicted that SaaS (let alone cloud computing) would be a thing, that large, profitable middlemen will directly host the software, pass through just the API, with zero customer exposure to the running instance…
> Nobody at that time knew or predicted that SaaS (let alone cloud computing) would be a thing I'm not old enough or experienced enough with computer history to seriously argue about this, but this statement sets off some alarm bells for me. My understanding is that extremely early on in computing history, especially early in the Internet era, people were thinking about SaaS. We had shared terminals, Sun was even try…
If you think about it the usage is best described tied to a an era of pay as you go services launched in the first decade of the 2000's.
There's otherwise not that much in common between Google Docs, pay-as-you go VPS/VM, Kubernetes, platforms like GCP/AWS, and Dropbox.
Re: AWS announces forks of Elasticsearch and Kibana
#629Earlier quoted context omitted.
AGPL also harms adoption at companies because of the "invasive" nature of the license. Google, for example, bans the use of any AGPL software. It's too vague for the legal team to confidently understand the "bounds" of the copyleft license. Permissive licenses like MIT/Apache have historically been what companies prefer to have when evaluating free software. And, generally, it hasn't been a problem for companies like…
None of this is new. GNU was founded and developed its licenses to address this issue. But there’s a tradeoff. One that GNU happily accepts. But there has never been a “code of honor” in this space. Give me a break. Companies have been offering Apache- and GPL- licensed software as a service for longer than Elastic has been a company. Elastic wanted to have its cake and eat it too. They overplayed their hand here and…
Suppose Elastic had used GPL instead. Then Amazon's hands are tied to a greater extent, but we still lose out. If AWS is the one maintaining ElasticSearch, then ES will steadily become more and more nichely-suited to being a part of the Amazon cloud and less suited to a self-hosted use case. Users can patch it for their own use-cases, but they'll never be able to make it as good as it would have been if there were full-time maintainers supporting them. The commons still become less rich.
The dream is that Elastic can keep maintaining ES but ES remains Apache2. Given AWS's presence in the market, it seems like that's not an especially viable option. For Elastic to keep maintaining ES while switching ES to a more restrictive license which doesn't prevent normal use cases but hamstrings AWS is less desirable, but still preferable to Elastic folding, because ES remains a tool which we can all use and patch as needed to power our own services.
Re: AWS announces forks of Elasticsearch and Kibana
#630Earlier quoted context omitted.
Slightly off topic but... > …except Elastic doesn't want to use the AGPL because anyone using Elasticsearch would have to open-source their whole codebase — a non-starter for most companies. They wouldn't have to. This is a common misconception about AGPL.
Is the AGPL not a copyleft license? What are the derivative works you're required to distribute as AGPL?