Why IBM doesn’t agree with Google’s Open Usage Commons for the Istio project
developer.ibm.com
developer.ibm.com
The other two projects under that are Angular and Gerrit, two other Google products.
IBM's disappointment stems from the fact that they feel Istio is also their baby and that it was suppose to goto the CNCF one day, not Googles own umbrella foundation.
I think the same. Google for the most part want their names associated with these products rather than being something community driven.
Maybe I'm cynical, but it seems pretty obvious to me that Google embraces OSS with Istio, Kubernetes, etc only out of necessity and being a distant 3rd in the Cloud race, and this is their attempt at some sort of differentiator.
With all discourse on the internet IMO there's way too much all one thing or another. Companies and people made lots of decisions surrounding various projects, and the reasons can be altruistic one day, about profit the next, or a little of both with the same decision.
When a company open sources something, you're right, they "see the value of open source". They know it will enable collaboration, essentially unpaid labor on their project, community and interest, and while others can use it, it places them at the center in a driving power position over the technology and gives them the ability to set the standard.
Open sourcing is good business. And that's before we talk about the marketing value of claiming it's altruistic and selling that idea to everyone else. Once you start taking advantage of people's naivety to promote your company and your brand, you're really cooking with gas.
People need to stop believing in corporations, because corporations are built for the purpose of making money. You can trust them insofar as you understand that goal and understand them to move in accordance with it.
And having worked on Google OSS projects, the 'unpaid labor' aspect often doesn't materialize. Looking at GWT , J2CL, Closure Compiler for example, 99% of all commits are from Googlers, very few outside contributions, despite for example, a peak of 150k MAUs for GWT in its heyday.
Google has 10s of thousands of OSS projects, only a small fraction of them are related to any kind of strategy. Most are people scratching an itch and throwing a codebomb over the wall onto GH.
If Google didn't think 20% projects provided value in terms of employee satisfaction and retention, or didn't find some value in the ability to potentially build on some of that random unplanned R&D if someone struck gold, Google wouldn't do it.
It's possible to have altruistic line employees, but corporations are always acting on cost/benefit. That benefit just doesn't always manifest directly into cash. Employee retention, good PR, etc. are all benefits that can be to some degree valued and measured.
Corporations do not have 100% visibility into what their employees are doing in totality and they often don't write any rules for or against something until after an employee does something so bad it creates negative PR.
There's no employee handbook rule that tells me I can comment on HackerNews, or tells me that I can't (well, I hope not). So until I get a call from Google HR telling me I did something bad on HN, I do it.
I don't think this is Google specific, I don't think the vast majority of employees in the tech sector obtain permission for both after hours contributions they make on the internet, or even stuff they do while at work, even though most companies have a policy not to use work computers for personal things.
People just don't compartmentalize their lives like that. Like say you're at the office, you're browsing some GitHub repo on a project completely unrelated to your company, and you notice an easy to fix bug, and send in a PR on your work desktop while at the office. Is your company allowing you to do this explicitly, or do they just not care, as long as you're being responsible and not spending all day at the office working on non-work stuff and not deliverables.
The COVID era brings this even more into focus, with everyone working from home now, really, how many people are doing other work during office hours?
You're trying too hard to attribute master planning to management, when the truth is, most of these tech companies are more chaotic and autonomous internally then most people realize.
Let's say you're the maintainer of the `ls` command. Some Facebook employee comes along and fixes a bug, out of the goodness of his heart. You're going to think, wow, Facebook is helping me, now I trust Facebook more.
Problem is incentives. If Facebook gets large enough, its employees who are trying to stake a claim in open source might not be thinking about how that ties into the company as a whole, if leadership isn't rewarding it.
It was (for better and for worse) the #1 reason to adopt GCP for our main cloud platform, and we're not alone.
It's has seen immense adoption that it wouldn't have had if it hadn't been in a neutral foundation. The work was definitely not done by Google alone (~30%).
IS GCP "still" #3 ? Yes. Is it because they didn't get a tighter grip on k8s and try to use it as a competitive advantage ? absolutely not. Nobody would have touched k8s with a 10 foot pole in that situation.
The strength of k8s is in its eco-system and that wouldn't have happened without the CNCF.
And I'm saying this as an early adopter of k8s, it's insane for the Google brass to see anything other than an immense success in the handling of k8s, at least from the perspective of how much money it made Google (indirectly)
It's not because of K8S strategy that GCP is #3. It's just because MS & AWS are better. PERIOD. Better at service, better at marketing, better at some product design sometimes, better at abusing their current dominant position (hehe it sucks doesn't Goog ? heh? Doesn't it ?)
I think you underestimate the revenue OpenShift is generating... If I see the support contract bills of my clients, holy crap, you could host a ton of stuff on GCP for that.
If I see the support contract bills of my clients,
holy crap, you could host a ton of stuff on GCP
for that.
on a more serious note: in order to compare apples to apples you should compare red hat support bills with GCP support bills (i think they call that "professional services"). I guess they aren't cheap either.
Sadly, also my experience. Response SLA of 2 hours? Please attach sosreport is ALWAYS the first reply. Even if you attached it in the original ticket (which we always do). The technical quality of the support is sub-par, most of the time when we include technical information, they don't seem to understand it at all...
Comparing consulting bills to IaaS bills is apples/oranges. Consultants are really just Humans as a Service (HaaS). A lot of the work we do is to bring in expertise in something that would otherwise be difficult and expensive to hire for and augment a team for a short period. We actually help customers with cloud-stuff (like AWS and Azure) all the time.
A lot of the work I've been doing lately is stuff that would have to be done if you were using K8s on cloud instead of OpenShift. I'm currently helping to migrate services from older platforms (like BlueMix) to OpenShift. You can look for that "feature" all day long in the cloud console, but you won't find it :-)
I'm also a freelance consultant mainly focused on OpenShift and k8s, my bill is separate.
Just wanted to say this one weirdly petty/obnoxious concluding line tainted an otherwise solid comment. Seems unnecessary.
i should check if it's still the case, but I remember a time where every effing kubernetes tutorial would make you spin up a k8s cluster on google cloud.
I'm trying to look up how to make my own cluster, not use someone else's. Figured it out eventually but it took way more searching then for learning how to use k8s.
Android brought people into Google's ecosystem.
Kubernetes could have positioned Google better in the cloud space but I think they were late to the party due to their failure to understand what customers want.
Summer of Code is a huge project they finance, seems altruistic but is a recruiting&marketing tool.
Publishing libraries like protobuf and abseil serves to drive standards in directions they like.
Chrome/Chromium drives the open web, ensures their say on standards and gives the ability to set them.
Android/AOSP brings leadership in mobile devices to them. Again driving standards to their needs and preventing Apple from making the mobile world a google-free walked garden.
SDKs for their services - obvious.
And to be clear, that is by no means bad. And that potential abuse of their position has to be controlled (i.e. when setting web standards to their likings) is mostly independent from their software being open or not.
Could you elaborate on why they care? Do they have some kind of competitive advantage when it comes to Protobuf?
With things like protobuf/grcp it's a bit harder, but if they define the standard for data exchange between machines they can eventually - use it as apart of their search indexer - incorporate external services into their products while keeping their standard architecture - integrate acquired software into their stack (you hope being acquired by google? Prepare ;) ) - it reduces risk of alternative approaches from becoming defaults, which could harm Google's integration wishes - if job candidates are used to protobuf there is one thing less to reach a new hire
This certainly isn't a conclusive list, and I can't judge relevance of each for their decision, but those things certainly impact the decision aside from "we think it's cool"
IME (I worked there for a long time, but no longer), individual (and small groups of) engineers with a passion for OSS drive publishing/sharing.
Execs with the interests you outline approve/agree to the sharing.
The motivations are not monolithic, and a corporation is not a person, it's a group of people.
Open-source should be extensively distributed over many actors to prevent any government censorship and attacks by political reasons.
It's noticable through their documentaion and in some of their architectural decisions and general approach/attitude towards software engineering.
I've have noticed that it's faded somewhat but it's still there, it's noticable specially when contrasting it with Amazon's approach which has a different approach or style to it.
I think it has to do with the origins of older google engineers (this also would explain why/how it's diluted but still there)
Also it seems that in some parts of google are trying to move away from these in order to be seem more enterprise friendly?
But personally, I really like google's more engineer-oriented attitude (here are they pieces and how they are made, you can figure it out) versus a more hand-holding style which later makes it difficult to move beyond the happy path.
it's not very easy to explain what I mean exactly as it all comes from a rather vague sensation about engineering style.
I don't think this matters to practitioners, it's a solution to fix trademark issues.
Open Source Licenses (like Apache under which istio is released) doesn't solve the trademark (TM) issue, they are vague around this topic, so the use of the name/logo/mascot to release any sort of managed or supported solution based on an OSS tool becomes a hurdle, a lot of money is at stacks here (imagine if k8s was still under the control of Google, AWS or Azure will never be able to launch something called Kubernetes engine because there is a risk Google will sue them).
The problem with CNCF and why istio was not transferred to that org is that CNCF acts as a mediator, so any decision around trademark takes soo much time and effort because CNCF has to get the blessing of everyone or build something neutral (like the k8s certification), and it's in their best interest that things are complicated as they can make money out of that.
The OUC was created to solve the TM issue and allow decisions to move fast, the usage guidelines haven't changed and will likely not change.
I do have to admit that myself i don't understand why no one from IBM is in the initial board of OUC
> IBM continues to believe that the best way to manage key open source projects such as Istio is with true open governance, under the auspices of a reputable organization with a level playing field for all contributors, transparency for users, and vendor-neutral management of the license and trademarks.
Based on this comment it sounds like IBM is saying that Google's OUC is not a vendor-neutral manager of license and trademark. It's also suggesting that there won't be a level playing field or transparency.
Basically IBM just wants it managed by an independent un-biased entity. Which sounds fair, if the whole purpose of the thing is not profit for one org, but the profit of a community.
More evidence to my argument that it is impossible for a corporation to own and develop a true community-led solution. The corporation is a slave to its shareholders and profit margins, so the community's wishes will never take priority.
Sorry, but that's the nature of standardisation, and the price you pay if you want to show you're really vendor-neutral. For me, actions like these make Istio go from the almost de-facto default choice for a service mesh, to suddenly wanting to consider other stuff.
> and it's in their best interest that things are complicated as they can make money out of that.
I don't really see how money would be an issue here? CNCF is a non-profit, Google is not.
That's not what the Cloud industry is about, it's about speed and time to market.
> and it's in their best interest that things are complicated as they can make money out of that.
Valid point, i have no evidence to support this, i just feel like CNCF looks for any opportunity to monetise itself for reasons i don't understand (Certifications, Conferences...). For a non profit that doesn't hire the core developers of the tools they support (most are hire by tech companies) i don't see why kubecon should cost 1k$+
Not sure how that's even an argument? Istio already was the de-facto standard, it is already in the market right now. Nothing prevented them from submitting to the CNCF - even if the acceptance would have taken a while. This wouldn't have suddenly stopped adoption or development in any way? It would only have enforced the view that the project was committed to stay vendor-neutral. Now, at least in my eyes, they've shown exactly the opposite.
> For a non profit that doesn't hire the core developers
Not sure why an organization formed to allow multiple vendors to work together on the same tools and play neutral party there would need developers. That would make them a stakeholder, something you want to avoid. They do have staff though[1], which also costs a lot of money, and organizing events isn't free.
Care to supply some evidences for such claim?
I know non-profit in the US has to fill public tax returns, i will be interested in figuring out how much execs earn at CNCF!
I run a conference myself and we manage to be 3 times cheaper with only 8 sponsors and we take care of travel and lodging for speakers. With the amount of sponsors they get, they could get away with $10 tickets like PyCon Europe which is free for all and still take care of travel and lodging for speakers. When Google gave Kubernetes for free to CNCF non profit, I don't think this is what they had in mind and may be why they went with OUC instead.
It's very expensive to run 10,000 person events, we run the events as pretty much break even across different projects, some events make money, some loss money, it all balances out. We also produce transparency reports for our events: https://www.cncf.io/blog/2020/01/09/kubecon-cloudnativecon-n...
Also keep in mind that nothing is free, running VENDOR NEUTRAL events is a hard business, ORM laid off all its events team recently during the pandemic (https://www.businessinsider.com/oreilly-media-layoffs-events...), I'm proud to say we haven't laid anyone off.
This article paints a more accurate picture of the situation where Google broke its promises to its partners, you should ask them why they did that: https://www.theregister.com/2020/07/09/ibm_oracle_cncf_prote...
Also, I believe people have been asking Istio to develop a conformance program for years and moving it to CNCF which had a track record of building one of the most successful open source conformance programs seemed reasonable: https://github.com/cncf/k8s-conformance
If not CNCF, there are other fantastic open source foundations like the ASF, EF, OSI, SPI and so on that could have helped.
This is a cynical point of view. At this point, Istio takes advantage of a number of IBM, Oracle, Lyft and Red Hat developers while not giving them any governance over it..?
"As one of the founders of the project, Google is the current owner of the Istio trademark. While anyone who is using the software in accordance with the license can use the trademarks, the historic ownership has caused some confusion and uncertainty about who can use the name and how, and at times this confusion has been a barrier to community growth. So today, as part of Istio’s continued commitment to openness, Google is announcing that the Istio trademarks will be transferred to a new organization, the Open Usage Commons, to provide neutral, independent oversight of the marks. [...]"
Here we have the interest of IBM/Redhat that as a services company benefits from an as open as possible model, whereas Google as a product company notices that their hugely successful k8s play does not turn into business for their cloud product as intended.
[1] https://twitter.com/ThatMightBePaul/status/12798801693443358...
Is k8s an amazing force multiplier or an unmanageable white elephant? Can it be both?
With Java and .NET, or now serverless middleware, I never felt the need for something like k8s, as there were already solutions to deploy a bunch of EAR/WAR/DLLs into some cloud based server, including having cloud based DB servers like Amazon RDS.
Taken to an extreme, I don't care where and how those packages are executing, k8, jails, containers, bare metal,...
The harder part is when there are byzantine libraries that tether deep into the OS. Often this is the case for some .Net on windows (and even with Java), mostly around commercial licensing models.
Spent the past two years pushing for getting (most) of the applications being worked on to at least be able to containerize and two major application developments are now targeting kubernetes for deployment. It's been uphill to say the least.
The only ones that I have had such issues with OS integration were anyway in desktop context, which were deployed via VM snapshots, for access via RDS.
All of these are pretty pricey.
Moving away from bespoke applications deployments has helped a lot.
So agreed that K8s is not Google's crown jewels by any stretch. They were inspired by the Borg paper I'm guessing though the early scheduler in K8s was laughable IIRC.
This thread is a pretty good explanation of how k8s is derived from the need to be compatible with some Borg concepts, while not deriving from Borg itself.
When GCP's container engine launched[1] in 2014 it was initially referred to by the anagram 'GKE' yet corresponded to the name 'Google Container Engine'.
After achieving CNCF certification in Nov 2017[2], GCP began using the term 'Kubernetes' within the product name, rebranding it as 'Google Kubernetes Engine'.
At least as far back as Oct 2017, members of the CNCF did not have to pay any fees[3] to participate in the conformance program.
Is it possible that executives within Google were unhappy about the marketshare gains of GKE and attribute this partially to the time it took their product to achieve certification?
If so, could that have led them to push other open source initiatives within the company to retain their trademarks under tighter control, even if that required creating an organization with less experience handling international trademark registration?
As per the post, IBM's stated standpoint is: "IBM continues to believe that the best way to manage key open source projects such as Istio is with true open governance, under the auspices of a reputable organization with a level playing field for all contributors, transparency for users, and vendor-neutral management of the license and trademarks."
[1] - https://www.infoq.com/news/2014/11/google-cloud-container-en...
[2] - https://cloud.google.com/blog/products/gcp/introducing-certi...
[3] - https://github.com/cncf/k8s-conformance/commit/34683a3316473...
After _creating_ the certification program: it was driven by GKE PM (and upcoming book author[1]) William Denniss. https://kubernetes.io/blog/2017/10/software-conformance-cert...
(At the time of GKE's launch, 'containers' were more well known than 'Kubernetes'. Google also owned the trademark, so it could have been called Kubernets Engine from the beginning.)
[1] https://www.manning.com/books/getting-started-with-kubernete...
PS: For anyone curious about the adoption of the word 'Kubernetes', Google Trends provides a good summary: https://trends.google.com/trends/explore?date=all&q=kubernet... (2014 was definitely very early days)
My understanding: Typically the trademarks are not covered by the open source license. If the software is licensed under Apache license, I can use the software to power my cloud business, but I might not be entitled to use the project's trademark in my marketing.
If enough people are upset about it, they can fork the project and take it in their own direction under their own governance.
To me, this is a non-issue. The people doing the work on the project are setting the projects goals and rules, and if you want to do work on the project, nothing stops you, either under their governance/name, or your own.
https://techcrunch.com/2020/07/08/google-launches-the-open-u...
And their FAQ: https://openusage.org/faq/#what-is-the-open-usage-commons
We know that. The confusing part is figuring out what does "managing open-source trademarks" have to do with other companies and developers...
IBM collaborated with Google on Istio. Google just handed the control over the Istio trademark to its Open Usage Commons group without notifying the Istio community. IBM is understandably unhappy with that.
disclaimer: work for IBM, but not on istio.
[1] https://techcrunch.com/2020/07/08/google-launches-the-open-u...
If this is true, the authors could be sued for breach of contract (what the damages would be, I'm not sure). However, I suspect that this agreement was verbal or implicit and wasn't actually written anywhere.
As they say, a verbal agreement is worth the paper it's printed on.
Get your promises in writing, folks.