Google donates money to help run the Kubernetes infrastructure | Hacker News Reader

Google donates money to help run the Kubernetes infrastructure

techcrunch.com

152 pointscoloneltcb51 comments

Disclosure: I work on Google Cloud.

I think the headline confuses the issue. The community wanted more input and ability to control the release and testing process for Kubernetes. It was a fair question to say "Why can only Google cut a release?". We're still happy to fund it, but to make Kubernetes more fully owned by the community and not Google, this was an important step.

Edit: Fwiw, I think William's blog post (https://cloud.google.com/blog/products/gcp/google-cloud-gran...) is more informative, since the TechCrunch article just quotes from it.

The headline is definitely confusing. I thought it was implying that Google was no longer going to use Kubernetes at all. The first couple of sentences of the article cleared that up, but the title should've been something like "Google hands off Kubernetes development to open source community."

Even that's not quite right :).

"Google opens up the management of Kubernetes testing and releases with the community" (or something).

I think this becomes very evident if you pay attention to the multitude of public Community Meeting (and various other SIG meeting videos). The community has wanted equal standing from the very beginning, and even if Google was stepping down funding, I think it would be an important step in making Kubernetes a fairer community, with no particular company seeming to have more input.

In these instances, can the URL of this post be changed to the primary source rather than reporting on it?

When will Google Cloud get IPv6 support to the VM? I went to a Kubernetes talk here in Seattle at Google's Fremont office, and I was shocked when I found out that IPv6 support was limited to the proprietary load balancers Google offers.

This kind of missing functionality makes Google Cloud dead on arrival for anyone needing low latency connectivity to customers on T-Mobile, Verizon and any other cellular network that only provides direct connectivity via IPv6. Natted IPv4 gets much shorter state table entries, and worse yet you get a latency penalty as legacy IPv4 traffic gets encapsulated in IPv6 on most LTE networks.

So you use the load-balancing service. That's what it's for.

That can be a bandaid but leaves you pretty far from being able to do traditional Internet style networking with global addressing.

I can't do that for SIP or WebRTC :c

Disclosure: I work at Google on Kubernetes CI / Infra etc.

The tooling, configs, etc are all open source [1] (and largely run on Kubernetes!), but the actual deployments have been Google operated. This is about opening up access to operation of the deployments. I'm not about to stop working on the automation etc. :-)

[1] https://github.com/kubernetes/test-infra

Edit: if anyone is interested in more, we have a post out on the Kubernetes blog today about some of the infrastructure: https://kubernetes.io/blog/2018/08/29/the-machines-can-do-th...

It's really great that Google people are coming forward to clarify it but what's with the scare-disclosurinating? Your affiliation is not a source of potential conflict of interest with some detail of what you have to say. It's the most relevant bit and the reason you can make a knowledgable comment. Say it loud, say it proud, etc...

I just want to be transparent. Everyone has an agenda, I don't want to be perceived as "sneaky" (have been accused in the past, even changed my username)

Yeah that is great and you should say it, when it matters. It's just that none of you are journalists who are write regularly and are expected by their audience to be reasonably neutral and who then disclose potential random conflicts in a standardized parenthetical.

You can just say 'I work at Google on whatnot and here is what I know/think/etc'. Lighthearted Suggestion: There's no need at all to write like HK-47 or Dwight Schrute.

Appreciate the feedback. From my end, I have been both accused of name-dropping (ooh, look at Mr special, he works at Google!) and of having an unstated agenda (hey, he's a phony, he works there!).

I found that simply stating is the compromise. I am not using my employer.ent to justify my opinion, just making my alignment clear.

And in fact, I align with Google AND with Kubernetes.

Yep, I understand it's fraught and as I've just amply demonstrated, someone is always going to whine about something. It's just weird to see you all line up one after the other and and robotically announce your affiliation in the format typically used for unexpected-conflict-of-interest when it's actually important, not a conflict and you're trying to be helpful.

That and I enjoy the largely futile but infinitely noble fight against various HN-mandated atrocities against basic English style and I want you to be free.

Echoing what Tim said, it seems to be standard practice on HN, I'm just trying to be transparent rather than somehow get accused of something(?)

I normally lurk more than I comment, but I'm disappointed to see how it's been represented. My teammates work hard _with_ the community.

Yes, I replied to Tim. I'm suggesting it's a dumb, apish standard practice. Write like the human beings that you are! "I work at Google, such and such happened".

FYI this headline is wrong is in now way is Google "stepping back" - this is Google becoming more open/transparent with how they traditionally handled CI/CD for the Kubernetes project.

Disclosure: I help run CNCF.io

Since people are complaining about the title, I took a crack at making it less misleading. If anyone can suggest a better (i.e. more accurate and neutral) title, we can change it again. It's a bit hard for an outsider to read exactly what happened here, or why it matters.

That's better. I really see this as "Google opens administration of Kubernetes infra, donates $9M in credits to keep it running".

We could also just link to Google's blog.

https://cloud.google.com/blog/products/gcp/google-cloud-gran...

Yeah the title is misleading, makes it sound like Google is abandoning the project

Disclosure: I worked on this grant.

TL;DR is that CNCF owns kubernetes, but until now I could not include non-Google people in the administration of (ostensibly) community owned stuff (google.com security policies in GCP). Now I can.

Google's involvement and commitment is in no way diminished, this is just a bad headline.

I've not been an avid reader of TechCrunch in the past, but there seems to be a trend of poor, baity headlines (and often story angles) at least among articles that make it to HN. Not sure if they've always been this way, if I'm reading the wrong articles, or if they're another casualty of declining integrity across the industry.

It’s too late now. All existing roles are now filled with Google people, and therefore the project culture has become the Google culture. No non-Google newcomers can individually change that culture.

So, this is unrelated, but since there are some googlers here - does Google really pay mediocre senior developers $250,000, on average?

Define mediocre. Google's hiring process is great but not perfect. Someone can always slip through the cracks in any company. As for compensation, Google can't differentiate who is worth 250K and who isn't solely based on the hiring process. A "mediocre" engineer getting paid 250K will likely not get promoted within the proper time and could be forced to leave.

yes

Anyone got a gdpr;dr for those of us who don't want to click ok on their cookiewall?

Google is donating $9 million in Google cloud credits to the CNCF (foundation that runs k8s) for the testing and operational bits. This means the CI and unit testing infra that was manually hosted by google will be in the control of the foundation and not strictly google anymore. The $9 million in cloud credits are expected to be used over the next 3 years.

Excuse my ignorance, but What???

How does one spend $9 million in Cloud computing by just running CI operations and Unit Testing in 3 years????

$9 million/3 years for Dev Ops Infrastructure for just one specific product sounds out of this world to me.

Run 5000 VMs more or less continuously. Spin up/down hundreds of kube clusters every day - on every GitHub Pull Request. Serve 9-digits worth of container images per month. To name a few.

The infra behind k8s is kind of staggering.

From the Google post:

> We’re funding this move with a $9 million grant of GCP credits to the CNCF, split over three years, to cover infrastructure costs. In addition to the world-wide network and storage capacity required to serve all those container downloads, a large part of this grant will be dedicated to funding scalability testing, which regularly runs 150,000 containers across 5,000 virtual machines.

> Between the testing infrastructure and hosting container downloads, the Kubernetes project regularly runs over 150,000 containers on 5,000 virtual machines, so the cost of running these systems quickly adds up. The Kubernetes container registry served almost 130 million downloads since the launch of the project.

It sounds like the k8s project has behaved as if it has infinite cloud resources at its disposal, and as such has made no effort to conserve usage.

If this really is a matter of giving the community a mountain of credits, they could probably make it last a whole lot longer with a quick housekeeping pass.

This is explicitly part of the process of transferring each property. We need to be able to produce reports on spend in each area, and those reports are property of CNCF, not Google.

Ok?

All I'm suggesting is that there are probably low-hanging fruit to make the credits last longer, if up until now the community has been behaving as if resources are free and unlimited.

I'm agreeing and saying that is part of the plan :)

Google tired of fronting the bill for k8s dev infra, welcomes CNCF to the neighborhood with $9m housewarming gift to maintain.

That's hyperbolic. Part of the announcement is a commitment to re-evaluate the numbers as time goes on.

Google is in No Way stepping back from this.

Uh I tried to summarize in the same tone as the article. That’s kind of the point of a tldr. I agree the article was clickbait and it’s obvious Google wouldn’t abandon kubernetes.

Thanks for the downvotes though.

When Google gives projects back to the comunity like this it usually also means that they will stop investing so many internal resources on them, and will move on to the next product.

We gave this project to CNCF years ago, and Google is still pretty heavily invested in Kubernetes. :)

This is the end of the last vestige of exclusive control.

> Google doesn’t say so outright, but it’s fair to assume that it wanted others to shoulder some of the burdens of running the Kubernetes infrastructure, too.

Might be a far fetched, but this might play together with the Commons Clause discussion. Nobody wants to do all the work without being compensated, while others make profits with the work.

I hope the transfer of projects to foundations and other open forums continues, and they are backed by industry leaders making the profits.

It should even go as far as finally paying solo-developers through some fairly devided pool of funds. Everybody wins when libraries are free, openly developed, production ready _and_ financially backed for LTS.

Reply on news.ycombinator.com