The Devops Fallacy

#devops #cloud #SDN On that which is seen and that which is not seen ...

When it comes to talking IT operations and financial considerations I tend to stay away from deep economic theories. I'm not Joe Weinman, after all.

But I happened upon (no, I don't recall how so don't even ask. The Internets, you see) an 1850s essay on political economics written by Frédéric Bastiat which used an analogy as the basis to explain his theory. Analogies are great because they're like pictures for grown ups and sometimes, pictures are necessary.

In any case, this particular essay is often referred to as the Glazier's Fallacy (also known as the Parable of the Broken Window) and the story focuses on a broken window, 6 francs, and a whole lot of economic theory. What captured my attention was a nugget in the parable that applies fairly directly to operations and in particular the business value of devops. Bastiat argues that, despite the silver-lining thought that says "oh, well, a broken window is bad but at least the glazier can stay in business", the broken window is actually bad because it prevents money from being spent elsewhere (and ultimately encouraging more economic opportunity).

Let us take a view of industry in general, as affected by this circumstance. The window being broken, the glazier's trade is encouraged to the amount of six francs: this is that which is seen.

If the window had not been broken, the shoemaker's trade (or some other) would have been encouraged to the amount of six francs: this is that which is not seen.

And if that which is not seen is taken into consideration, because it is a negative fact, as well as that which is seen, because it is a positive fact, it will be understood that neither industry in general, nor the sum total of national labour, is affected, whether windows are broken or not.

Now let us consider James B. [the shopkeeper whose window has been broken] himself. In the former supposition, that of the window being broken, he spends six francs, and has neither more nor less than he had before, the enjoyment of a window.

In the second, where we suppose the window not to have been broken, he would have spent six francs on shoes, and would have had at the same time the enjoyment of a pair of shoes and of a window.

Now, as James B. forms a part of society, we must come to the conclusion, that, taking it altogether, and making an estimate of its enjoyments and its labours, it has lost the value of the broken window.

Ignoring the politics, if we apply this same parable to operations and a misconfigured server (as opposed to a broken window) we start to see the value of not having to spend time fixing things that are broken. "Now, as James B forms a part of operations, we must come to the conclusion, that, taking it altogether, and making an estimate of its value and its labor, operations has lost the value of the misconfigured server."

In other words, the economic case for devops is based partly upon the reality that time spent fixing things is lost. It's a negative; it's not just that we gain the time when devops is applied and deployment lifecycles are made successfully repeatable. It's that we also gain what we had lost spending time tracking down errors and fixing them. "Enjoyment of the shoes and the window" in operations equates to "enjoyment of new value and a properly working server."

In other words, it's nearly a double gain for operations because that time that was spent fixing things is now spent on adding value and is not lost in troubleshooting.The value of devops is computed not just by the value it can add, but by continued value of the server working as expected.

That which is seen (the server) and that which is not seen (the new value that could be added were operations free to innovate).

We generally articulate the value of devops by saying "we'll have more time to be more responsive or innovate new services" but we forget to add the value of that server that continues to work as promised while we're innovating. That value remains and it actually is a positive gain because we aren't expensing time against it.

When we're trying to articulate the value of devops to the organization, we need to include both the sustained value of properly working systems as well as the new value added. Focusing on the positive impact and value to the business in terms of dollars and time (not always the same, as Bastiat theorizes) may help sway those still unconvinced of the value of devops.

And for those focusing (or starting to focus) on SDN, there's a similar argument regarding the positive gain of a more self-managing network in addition to new value added. Et tu, cloud. The general principle applies to all technology that enables systems to run smoothly on their own.

Food for thought if you're trying to justify getting a technology initiative like SDN, devops, or cloud funded and running into roadblocks.