As part of my self-imposed DevOps induction I’ve been casually following the #devops feed on Twitter, and a tweet from @emmanuel_belo got me thinking – seemingly enough to spur me into firing-up Live Writer!
Whilst the underlying point is rather self-evidently correct, I view it as a bit of a straw-man argument. There are a multitude of difficulties associated with delivering quality software of which giving the user or customer actually what they want is just facet (albeit an important one). As such there are plenty of strategies out there that aim to address this, but few of those really tackle the issue of the Development/Operations divide in any detailed way – for those that do, that part of the message tends to be lost amongst the rest of the goodness that they proffer us.
It’s all to easy to say that DevOps doesn’t solve the whole problem, represents nothing new, or is nothing more than common sense and therefore deem it to be of limited value.
Working in the consulting business as I do, does mean that I tend to miss out on the long-tail part of the application lifecycle – keeping it running in production, on-going releases etc. However, the flip-side of that coin is that I get to see a lot of variety in terms of the technology, business models, delivery practises, team dynamics, to name but a few.
The reality is that even in organisations that have done a good job at adopting these ‘other’ strategies for improving their software delivery capability, you still see this divide between their Development and Operations teams – no matter how agile or how refined their engineering practises are such a divide can still be hugely divisive.
It is on this premise that I see the value in the notion of DevOps; it allow us to bring focus to a particular set of problems and in doing so (whilst not primarily concerned with the end-user or customer) should help ensure that customer requirements, particularly non-functional ones, are far more likely to be met – which seems like a ‘good thing’™ to me.
To view DevOps as aiming to solve all our software delivery ills is to fall into the silver bullet trap, and frankly, miss the point.