It's not super clear why Red Hat's market cap of "only" $11bn means that their business model doesn't work. It would seem to me that it works pretty well -- to the tune of $11bn.
That is, Red Hat has basically won the enterprise datacenter Linux war. It's Red Hat Linux and Windows, at the cost of Solaris, AIX, HP-UX, SLES, and others.
And despite that crushing success, and having no worlds left to conquer in the OS space, their market cap is only $11b. Salesforce.com has a 12% market share, and they have a $38b market cap.
EDIT: just cleaned up "Linux" vs. "Red Hat Linux"
Solaris / AIX / HP-UX still get plenty of mileage in the legacy application (Oracle!) space. It's going to take the enterprise longer to migrate to systems like CoreOS, but RHEL doesn't have the kind of automatic lock-in that some of these legacy systems have.
As far as I can tell, RH is focusing revenue growth strategies on things like OpenStack, and similar product suites, so the company will survive. But by the time RHEL 7 comes out, what's it going to offer anyone in the datacenter? The days of the monolithic distro are numbered.
But speaking as someone who literally based their whole business on managing Oracle on *nix, Solaris/AIX/HP-UX are dead platforms. Over 80% of the Oracle servers our users are managing with our software are Linux. Of those, probably 80% of them are RHT, and 15% are Oracle Linux. SLES and CentOS barely register.
There is one more massive advantage that RHT offers vs. anyone like CoreOS - platform stability. Things like driver interfaces, library versioning, kernel versions all make it much easier for ISVs and corporate engineering teams to certify and manage RHEL releases. They're not a fast-moving target.
Over time that will become less relevant, I agree, but it's been a huge factor in their success.
I can't quite speak for HP-UX, but Solaris and AIX are definitely not dead platforms. My main claim is that "Big Iron" is going to outlive RedHat in the application platform category. Oracle the DB is one thing, but the sprawling ecosystem of non-database applications that Oracle has accumulated over the years aren't going to keep needing heavier distros. A lot of that stuff is moving to VM image instances anyway.
What a platform like CoreOS (or other ways of combining LXC with your hardware infrastructure) offers is separation of two major categories of concern. Driver interfaces and kernel versions are hardware sensitive. Application dependencies on specific library versions far less so.
Your vantage point makes sense, though. Databases are really the only application left in the enterprise that haven't seen this decoupling from hardware. But that doesn't bode well for the total number of Red Hat Linux installations.
> I can't quite speak for HP-UX, but Solaris and AIX are definitely not dead platforms. My main claim is that "Big Iron" is going to outlive RedHat in the application platform category.
I have a hard time seeing big iron becoming anything other than the open systems equivalent of legacy mainframe.
Bear in mind that kernel interfaces aren't purely about hardware - it's about repeatable, predictable performance and behavior profiles. Watch a bunch of bank server platform engineers lose their shit when memory performance under pressure changes after a kernel patch.
And remember that in the traditional enterprise, it's not just about stripping down the OS, it's about all of the other tools that get used for performance, security, backup, monitoring, remote execution, change management, privilege escalation. I actually am working on some research on things like LXC at my day job, and one of the challenges is that containers are great until you want to run a bunch of stuff inside the container.
And my vantage point was at one time databases, and that's still the market segment I know the best. But I work on everything server and cloud-related, so most customers I talk to are just traditional enterprises.
My company (15k+ servers) uses ubuntu.
Our customers are big traditional enteprises, by and large. Banks, telcos, manufacturing, oil and gas, etc. They love RHT, or if they don't love it, they at least appreciate the stability and ISV support.
VMware 1998 Microsoft 1975 Oracle 1977 Amazon 1994 RedHat 1993
At least two companies had a major head start and should be expected to be larger.
One doesn't sell software, it uses software (Amazon).
The remaining two are VMWare and Redhat and the success of either can be attributed to the birth of their markets.
The virtualization hardware mostly began with a commercial product and the commercial products were able to stay ahead by being better capitalized than any efforts to commoditize the product.
With the other market, server operating systems, Linux was already a community effort with several distros and you couldn't have exactly have even tried any other strategy than open source without completely alienating your early adopters and dying on the vine before getting to a minimum level of adopter.
That aside, I do like the point of the original article. Build a scalable service on top of something open source, rather than T&M services.
37Signals/Basecamp are an interesting example of this. Rather than building a business around supporting Ruby on Rails, they've built products on top of the framework they created. It's certainly an outlier and I can't think of many other examples like them who both built a popular tool, but then just used it as a tool instead of trying to monetize it directly.
DHH wrote: "External, expected rewards diminish the intrinsic motivation of the fundraising open-source contributor. It risks transporting a community of peers into a transactional terminal. And that buyer-seller frame detracts from the magic that is peer-collaborators.
It also holds the threat of corrupting the community at large. We plant the seeds of discontent by selective monetary rewards."
http://david.heinemeierhansson.com/2013/the-perils-of-mixing...
I've certainly seen this recently in some open-source communities. The peer-to-peer camaraderie transitions to a dictator-to-underlings relationship and people start defecting. It's sad to see, but happens when an open-source company isn't vigilant in nurturing the community that got them where they are.
Kudos to those companies that are able to pull off maintaining a strong community while still staying focused. It's a rare thing to be able to do well and behind the scenes there's a lot of work that goes into that kind of nurturing.