The Skeptical Informer 2013 number 7: DevOps, another 5% Club or the next iteration of ITIL?

The IT Skeptic's Skeptical Informer newsletter for 2nd April 2013.

The Skeptical Informer

Recently I have been getting involved in the DevOps world a little, to better understand it.

I read The Phoenix Project, which was worth reading. I reviewed it here.

I attended New Zealand's first DevOpsDay and presented - read on for more about my topic. It was an interesting experience that looked unlike any ITSM conference: MacBooks and T-shirts everywhere.

And I was a guest on DevOps Cafe with Damon Edwards and John Willis, talking about the topic of this newsletter - as i said, read on. Go listen to the DevOps Cafe after you're done with this newsletter - we expanded on this topic a lot.

(DevOps Cafe is a DevOps equivalent of our own ITSM Antipodean Podcast - latest edition hot off the press, talking with Scott Spence about Prince2 and training.)

The adherents of DevOps promote it with a passion that sometimes veers dangerously close to a cult. At least they don't speak in the tone adopted by some Lean proponents, who sound more like they should be cross-legged on top of a mountain:

Great work requires people to engage on 3 levels: Head (intellect), Hands (physical), & Heart (caring enough to take ownership of outcomes)...
When emotions are high, opinions (not facts) rule our decisions -better to take a time out & let things calm down before setting course!

...not to mention The One True Way

multiple models of improvement, behavior & problem solving, run the risk of "model confusion" and people get baffled.

No, DevOps folk are much more practical. But many of them still have an evangelical zeal, and a conviction of DevOps ability to fix everything and to be applicable everywhere, that is a bit alarming.

Devops works. We can see that in some examples. Just how many is not clear to me yet. Nor do I fully understand the context within which it can be expected to work, though I am forming my conclusions.

I see DevOps working best in systems which are relatively standalone, simple, and low risk, (not complex entangled poorly-understood multi-layered legacy systems). Systems that

  • can be developed by small dedicated team(s) of high-performance professionals
  • support "infrastructure as code", which rules out a lot of legacy systems
  • are amenable to automated testing
  • the users tolerate frequent incremental change.
  • the business can tolerate production errors so long as they have a short lifetime, e.g. NOT air traffic control or foreign exchange clearing. A movie rental website can afford to lose or screw up a few transactions occasionally.

It seems to me that two scenarios are likely:

1) Companies that can really get value from DevOps turn out to be another Five Percent Club, like adopters of CMDB. A small proportion of organisations will be passionate proponents of Devops because it worked for them. More will thrash around with it for a while before giving up or ending up with a half-assed adoption. Even more will do something they call DevOps but which isn't. And another bunch will never do anything. Then reality is that only 5% ever should have even attempted DevOps, after qualifying themselves against my checklist above. Just like CMDB. And just like CMDB, the analysts and vendors will jump on the Devops bandwagon, trying to convince everyone that Devops cures whatever ails you.

2) DevOps will be a major contributor to the next iteration of ITIL. i.e. DevOps has something to teach all of us. Just Like ITIL, DevOps is not an all-or-nothing packed product: it is a set of guidelines that one selectively adopts and adapts. As I have been hammering in my DevOps presentations, DevOps and ITIl are just different views of ITSM. The key concepts of DevOps will at the very least be seen as a credible option to be considered in all cases, or they will have an influence on how we think about doing IT.

The third scenario is that DevOps will sweep the world and become the default model and worldview for IT. I find this unlikely unless IT solves the legacy complexity problem. Once IT has a simple architecture of modular loosely-coupled small systems which provide discrete business functions to be assembled into self-managing services, only then will DevOps become the leading approach to creating and delivering services.

My money is on #2, hence I have started an initiative to find the common ground between DevOps and "trad ITSM" as personified by ITIL; to reconcile their worldviews; to extrapolate both to find a common model. I'm calling it Kamu (here's why).

kamu

One of the biggest problems with the whole DevOps-ITSM thing has been the religious positioning of DevOps as the antithesis to ITIL. Which it is, they are the two ends of a spectrum. But some voices have called for the death of ITIL, the overthrow of ITSM, the ascension of only DevOps. This has alienated the "Robust" camp who see DevOps as a barbarian attempt to pull down the defences that keep IT safe and stable.

And of course the IT Skeptic foams at the mouth at any attempt to position a new idea as the magic pixie-dust solution to complex problems. Which may have happened once or twice over Agile and DevOps :)

Both camps tend to compare their own perfect theory to their opponents' imperfect practices, and compare their own poster-child success stories to their opponents' horror stories. The same company names come up over and over ....

Lately more nuanced, reasoned positions are emerging, with rabid anti-ITIL voices like Galtieri and Chambers replaced in my feeds by the more moderate ones of Kim, Humble, Orzen, Willis...

I will devote 2013 to helping reconcile the world-views of ITSM and DevOps. Please join me with Kamu - there is a discussion group on Google+. There is a unified understanding in there somewhere.

brothers in arms
Image © Copyright canstockphoto.com


Reminder: go listen to DevOps Cafe with Damon Edwards and John Willis

My presentation on Kamu at DevOpsDay NZ

Some related posts from the blog (don't forget to check out the comments! Always good):

Kamu: a unified theory of IT management - reconciling DevOps and ITSM/ITIL.

A quest for a unified theory of IT management is not a flag of ITSM surrender. Some folk have interpreted my last post on A unified theory of IT management as surrender to the DevOps movement: "OMG DevOps was right all along, what a fool I have been". Not at all. I'm saying both the DevOps and the ITSM communities need to move on and find something that works for everyone.

The Five-Percent Club. The Five-Percent Club is that elite group of the (less than) 5% of organisations who actually succeed in justifying and implementing a CMDB, or the more modern and equally nutty CMS.




Please pass this newsletter on to those you think will find it interesting.
If someone sent you this, go here to subscribe to future editions.

Go here to comment on this newsletter or to see other editions of the Skeptical Informer.

Follow the IT Skeptic on twitter: @theitskeptic
Follow the IT Skeptic on Google+
You can now also follow the IT Skeptic on facebook - your Like is welcomed.

The IT Skeptic's blog
Blog RSS
Comments RSS
The IT Skeptic's ITIL News Feed

Syndicate content