Thursday, December 12, 2013

Ignore Linux Standards at Your Own Peril

by Dietrich Schmitz

So, I've been watching the progression of all things Linux for the past few months and noting how many Distributions are falling in line with systemd adoption and those which are not.

Systemd adoption by major Linux Distributions

The freedom to make choices in software design which introduce variation ultimately exacts a cost.  It's not apparent up front but in the long term, introducing 'variation' of any kind into a operating system results in additional complexity.  Only recently has holdout Debian gotten around to recognizing that their sysvinit system is beginning to show its age and 'limitations'.  Uniformity lends to conformity and implied standardization, sometimes in the form of 'de facto' standards.

Today, I am reminding readers of the importance of adhering to standards as the image below shows how being Linux Standard Base compliant can and will lead to lower cost of operation and ongoing maintenance.

Linux Foundation promulgated 'Linux Standard Base'

Does having to adhere to standards stifle creativity and freedom of choice?  

Well, there's room for debate on that question and while to a degree standards compliance does impose a restriction on the respective developer's freedom to deviate, it also results in reducing complexity and to a larger extent fosters an environment in which software vendors and business concerns can reliably make assumptions which utilize those standards, including standard software behaviors, to streamline cost of operation from both consistency and reliability points of view: less complexity, fewer points of failure.  With fewer points of failure, reliability goes up and support costs go down.

Is my preferred Distro 'X' Linux Standard Base compliant?

It's an important question.  Let's take a look at Microsoft Windows for an answer.

Historically, over the past fifteen years preferring to work with Windows has been rightfully perceived by CIOs and CEOs as an implied standard, and despite Microsoft being a Monopoly, every developer knows that they have only one software API to deal with when choosing to code Windows executables on the Windows Legacy x86 platform.  That makes Windows an implied or 'de facto' standard, in lieu of there being any other competitor in a market.  And the choices imposed on Developers are made simple.  There's just one theoretical 'Distro' to think about when it comes to designing a piece of software for a given market.

But, there aren't any 'Distros' in the Windows world.  

Of course, but you see, Distro variation is non-existent, along with the implicit complexity stemming from compiler dialectic variations, differing package management and file hierarchy structure design considerations.  They're all gone and the attendent costs exacted on program development don't pertain in the Windows software development thriving ecosystem.

Should there be fewer Linux Distros?  

I don't argue for fewer Distros.  However, I do maintain that if the designers of Distro X conform to a base standard and not deviate, they will not have compromised their right to creative choice.  One need only look to the richness and diversity of software written to the Windows API to have an answer.

As long as Linux developers continue to clone new Distros and cheerlead the way, all the while ignoring standards, the chance for long-term survival of their work is put at peril.

Big Business Craves Stability

Simply put, Big Business Enterprise cannot afford downtime and thus the reputation of any Enterprise grade software system hinges on its stability and reliability in addition to effectiveness in providing a business solution.

There will not be as many Distros in five years as there are today.  

That fact is a certainty.

By attrition, some Distros will simply fade away into disuse.  The sturdy long-termers will consolidate and survive by adopting common standards and as there will be far fewer Distros, more programming effort will be spread across the few remaining base Linux Distros.  My prediction is there will be no more than six remaining in five years.

But be assured, Linux will continue to endure and survive, flourishing on a base set of LSB-compliant Distributions for many years to come.  -- Dietrich
Enhanced by Zemanta


Post a Comment