The business case for DevOps automation

In the past I have written sceptically about automation http://www.itskeptic.org/automation.
I have learnt better now.

The business case for automation can seem questionable when the work being automated is a small effort or seldom causes errors. But it is only questionable if we have to narrow a focus on the scope of the benefits of the automation. If we think in terms of a local optimum then it may not look like a good investment but if we think in terms of accelerating and controlling the entire flow of work from into end then the benefits are much greater. Almost all automation can be justified when we look at the impact on the entire flow of value of work.

If you were automating the last 5% of a flow of value then it could cost almost anything and you could still make a case for it.

What are these benefits when we think systemically?
Once we have an automated flow we can iterate frequently, allowing us to become agile.
Even when we only automate one piece of the flow, it still empowers us to execute that task as often as we desire, for example performance testing a piece of code.
And of course as my friend BMK ( @lbmkrishna on Twitter) said at a recent client DevOps Showcase (DOSho) "If you automate to stop having 6 people work for 3 days across a weekend to deploy, the ROI is immediate"

What do you think? Any other benefits?

[A couple of great additional thoughts from the comments below:

Mustafa Kapadia
There is a lean principle that states, for every 25% decrease in cycle time productivity improves by 20%. We have used that for both task specific activities as well as across the entire SDLC.

Plus the real benefit of speed shows up on the business side, through kpi such as revenue, less churn, customer engagement etc. in my experience trying to contain the benefits to just IT budgets does not make sense...it should be across all entities. Is that what system thinking is all about .

Kevin Smith
Automation can be structured to consistently collect data about the execution of the process. That data can be mined to improve the process, as well as monitored to detect variances or failures. It can offer enhanced visibility into the system, providing decision makers, both human and automated, a better set of facts to reason from.

]

Syndicate content