Gould: Oracle to Red Hat: It's Not Your Father's Linux Market Anymore

No. In fairness, Red Hat's logos and other trademarks are proprietary. If you can strip the binary packages of those, you can redistribute the rest just fine. All of it is open source after all.

This is precisely why CentOS exists. Doing the above is hard. It would also require someone to have access to the binary updates via subscription. It is easier in the long term to strip the source of marks, rebuild and distribute that. Particularly because this allows CentOS to have its own branding and patches, where required.

Is Beyer brand aspirin still considered proprietary when there are generic aspirins on the market sharing shelf space but not sharing consumer brand recognition or loyalty? I'm not sure that ordinary people speaking ordinary language would consider it so. I certainly don't think of Beyer aspirin as proprietary. Beyer is certainly brand differentiated, and is priced at a premium compared to "generic" options...but proprietary?

-jef

1 : one that possesses, owns, or holds exclusive right to something; specifically : proprietor 1

2 : something that is used, produced, or marketed under exclusive legal right of the inventor or maker; specifically : a drug (as a patent medicine) that is protected by secrecy, patent, or copyright against free competition as to name, product, composition, or process of manufacture

Now, you can use the word proprietary anyway you want, but if your unusual definition of the word is the "main point of your article", than why are you so upset with people here for thinking that you sound like a "shill", or that you are spreading propaganda? Proprietary has a very specific meaning in the software world and you are clearly distorting that meaning in your defense, and yet your are also clearly using the term in the common way to evoke emotional responses.

No I am not going say that we are saints here at LWN, but I think that you will find that we have high standards for the articles which we expect to be linked to from here. Maybe you are not used to this? And perhaps the name calling against you was not called for, but at least it had a point and was related to the perceived criticism of your article. You, however have elevated the name calling to a level which is way beyond what most of us expect here. Perhaps you need a thicker skin to be a reporter, because while the readers here may have poignant attacks, they are rather milder than the general public who might have just called you an "idiot" as you so aptly did here.

To be honest, I think we all expected more professionalism of you, if that is the sort of comment response that you think will make LWN readers more respectful of your articles, it has likely failed miserably.

When you write stuff like this, you just show that you understand little. Red Hat's strength lies in a simple fact: they were willing to risk properly supporting a Linux distribution and they are very good at it.

How can RHEL be "carefully and strategically crafted" when it's essentially a branch of Fedora? The mind boggles...

From your summary:

> Like Google’s Android, it suggests that Linux is beginning to fragment in the same way that Unix did.

Er, sorry, I had a look at kernel.org (I didn't actually - just a rhetorical point) and haven't seen anything changed. Care to share something we don't know?

> While Red Hat may tolerate CentOS and other cloners (provided that they strip out Red Hat trademarks), it does not approve of customers who want to use genuine RHEL without paying for it.

You for real? You can use RHEL without paying on as many machines you like. You can copy it to death. Red Hat won't do anything about it.

What you cannot have is a free RHN account to get updates. That is what you pay for: maintenance and support.

What Oracle did with the kernel recently is something they need to run Oracle DB on top of it better. Good luck to them with that (not sarcastically - I promise).

Most Red Hat customers, however, pay Red Hat for the whole thing to stay essentially the same. Yeah, I know - it may sound stupid to you. But it's not. Red Hat guarantee binary compatibility within the same version of RHEL and security patches for the life of the product. No idea whether you have some software development experience, but it takes quite a bit of effort to backport stuff like this. At the same time, they enable new hardware with this "old" stuff, so if you are a big company, with lots of investment in the software you use, you get to keep it for longer. That's Red Hat's value proposition - maintenance and support of something that otherwise moves too fast.

They aggressively work with upstream developers and will often make sure that the improvements they ship are widely available to many different distributions and corporations well in advance to Redhat shipping anything. KVM, improvements to GCC, driver enhancements to the kernel, performance improvements, usability improvements, security improvements. Redhat developers a lot of features that end up in other distributions for a long time before they ever end up in Redhat's shipping product. Not always, but this is often the case.

This is what makes Linux's 'fragmentation' different from Unix. With Unix it was fragmented because of restrictions placed on use. Those optimizations and proprietary features are associated with very high costs. With Linux it's only fragmented because people are optimizing, each in their own way, to fill nitches and functionality that is unique for their own situations and desires. So-called 'proprietary features' have no cost associated with them other then the work you have to put into implementing them (which is very low because all the code is already written and debugged for you). It's almost all benefits. There is nothing that Redhat ships with their software that cannot be duplicated by anybody else with a little effort.

Oracle should know this better then anybody because Oracle created their own clone called 'Unbreakable Linux' based directly off of software downloaded from ftp.redhat.com. Although they did not do as a good as a job as CentOS does.

Redhat knows that their source code improvements, design choices, and their optimizations need to be done because they benefit their customers. Exclusivity for exclusivity sake has NO benefits for customers at all. Customers don't buy the Redhat OS because that is the only place they can get that OS or those features. (because it very obviously is not true). They buy it because Redhat works hard with a huge number of different people and corporations to provide the best possible software in the best possible way for their customers. Redhat has a large pool of knowledge and the connections and partnerships with all sorts of different industries that their customers find extremely valuable.

Support options, ISV certifications, government validations and certifications, approval processes in large corporations, hardware vendor support, etc etc. Willingness to work with all the GNU/FSF/Kernel.org/GCC/etc type folks in the world, on terms favorable and preferred to everybody involved. Those partnerships and relationships and the depth and breadth of knowledge that Redhat has about it's own OS and it's own code base is what is valuable. No body else offers that sort of value in a OS; short of Microsoft, and Microsoft's costs are exponentially higher then Redhat's.

Redhat does not 'tolerate' other people using their code... Redhat benefits heavily from the independent developers and other projects that use that code for different purposes. Projects like CentOS benefit Redhat because customers can use CentOS to keep deployment costs low without raising the cost of supporting those customers one bit. They can pick and choose and pay Redhat based on the value that Redhat's services and support can get them. This way customers can have the highest level of utility from the money that Redhat receives from them. This is true with almost all the software that Redhat ships.

To put it another way: There is no reason to pay Redhat money besides the fact that subscribing to Redhat's support contracts will improve your own organization and lower costs. If you want the features that Redhat offers then you do not have to pay a dime.

This is why Oracle will fail and will keep on failing if they think they can just make their own highly optimized version of Linux and think people will buy into that. It's just plain wrong-headed thinking.

The stuff your talking about has been tried many different times by many different Linux companies and it always is going to lead to higher development costs, longer development timelines, and less stable software. Oracle will have higher development costs and the complexity and size of the QA process necessary to compete with Redhat will be something that Redhat themselves do not have to deal with. Redhat thrives because they have been able to avoid that trap and have been willing to abandon exclusivity of code and OS features as a reason for people to pay them money.

Right now Linux is a highly optimized and mature server platform. There is very very little that Oracle can do that will make their special Linux version perform in ways that cannot be duplicated by anybody running Redhat, Debian, Ubuntu, or anything else. If Oracle can produce a packaged product that is easy to use and is cheaper then running Redhat on a Dell or other vendor, then sure they can make some money. But Oracle's track record is not hot in those regards.

Here is the question you need to answer: would Red Hat be just as successful should they drop their super-secret and carefully crafted development effort (Fedora) in the future and join (for instance) Debian with all of the resources and then build RHEL from it?

My answer to this question is: yes, of course. Yours appears to be no, given the article you've written. Or do you care to revise that?

PS. That "super-secret and carefully crafted" above made me chuckle. I mean the whole thing is so "strategic" that they let idiots like myself have complete visibility of it :-)

When you fix up your "admittedly garbled account" and misuse of a technical term to mean something else, maybe you can improve others understanding of what you were talking about. I would also suggesting that you don't use the word "proprietary" because it is a very loaded word considering the audience that will read your article. If you are keen on fixing up the article so that it uses the words accurately, we will have the time to reflect on the improved article. My understanding is just fine. Thank you very much.

Well, you got advertising hits, and that's the main thing I guess.

I have been paying fairly close attention to open source software development (as software and as a movement or phenomenon) for something approaching 2 decades, and I have rarely known the term "proprietary" to be used the way you define it, whether in a commercial or open source context, with a single categorical exception: by senior executives or leaders of major software firms, or their mouthpieces, in an attempt to obfuscate or minimize the differences between their offerings and a perceived competitor's offerings, or more recently and commonly, to obfuscate any perceived benefits to open source software.

Examples of people who have gone to great lengths to attempt to redefine the term proprietary, as well as other terms crucial to the identity and benefits of open source software, include Steve Jobs, Steve Ballmer, Bill Gates, Scott McNealy, and of course, Larry Ellison. Usually (uh, actually, always), these re-definitions of key open source terms or concepts are crafted in a way as to render the term meaningless as a way of distinguishing between two or more options. Your re-definition of "proprietary" does exactly that. I assure you that the vast majority of informed people would agree that to be proprietary there must be meaningful constraints on the reproduction of a product that go beyond the mere manual effort required to replicate some small portion of the total work that went into the creation of that final product. In this context such constraints include legal constraints, obfuscation, and limitations on the availability of the source code.

Given the above, you should not be surprised that you may be accused of serving as a shill for Oracle or Larry Ellision, even if that was not your intent. Either way, this element of your piece draws into serious question your qualifications as a Tech Analyst, your analytical capabilities, your objectivity, or any combination thereof.

> "differentiated with a view to achieving competitive advantage in the market (and possible some degree of customer lock-in)"

So, when Oracle copied RHEL, that copy was also proprietary. This time to Oracle, because it:

> "differentiated with a view to achieving competitive advantage in the market (and possible some degree of customer lock-in)"

So, these two identical copies are differentiating nicely and locking folks in. Because, you see, if you develop a program that runs under RHEL/Oracle Linux, it won't run anywhere else but on those two proprietary, locked in systems. OK?

Now, if you take the new Oracle Linux, with that modified kernel, in that case you cannot:

1. Port those patches to anything.

2. Run anything but Oracle software on it.

3. Run Red Hat proprietary software on it, such as OpenJDK/JBoss, unless you want the server to explode.

I can go all day with this nonsense... ;-)

That's because I'm not really a Red Hat insider. Never worked there. Just used the darn thing for a while now, like many here.

Since it depends on education and culture it is probably true in some communities. Depending again on your education/culture you could even be unable to think otherwise.