Biggest mistakes in DevOps? Trying to do too much, not prioritising and setting over-ambitious targets

Expert panelists at Computing's DevOps Summit 2017 divulge the biggest mistakes they've made in their journey to DevOps

Trying to do too much, too soon; a lack of prioritisation; and, setting wildly over-optimistic objectives and targets are just some of the mistakes that the expert panel at Computing's DevOps Summit 2017 warned delegates to avoid as they start their journey into DevOps.

Sandra Christie, practice lead, analysis, at retailer John Lewis, Ryan Bryers, chief technology officer at e-payment company Worldline UK & Ireland, Dale MacDonald, technical team lead at Thomson Reuters, and Tom Clark, head of commons platform at broadcaster ITV, were responding to a question from the audience, 'What's the biggest mistake you have made in your journey to DevOps?'

"I tried to do too much at once," said Clark. "I come from a development and engineering background, so it's almost like I was given this big trainset on which I could do whatever I wanted and suddenly, I wanted to do everything…

"It didn't really have an impact on the platform, because I had a great team, but I nearly ended up murdering myself over it.

"So I learned a lot about prioritisation; must-haves, could-haves, should-haves; and what should come first.

"I have a big 'Trello board' of all my tasks. I've still got tasks on there that I came up with three years ago, and I will get to them eventually, I keep telling my boss, but you have got to prioritise - you shouldn't try and do too much at once," said Clark.

McDonald at Thomson Reuters echoed Clark's advice. "There haven't been any train smashes, but one of the things [we grappled with] in the early days was getting the balance right. You've got all these new, fancy tools and gadgets. But one of the things that you really do need to measure is that you are actually making progress."

So, advised McDonald, it's important not just to devise the metrics for measuring progress, but to revisit them, three- or six-months down the line.

For Sandra Christie at John Lewis, it was her enthusiasm for ‘objectives and key results' (OKRs), a framework of defining and tracking objectives and their outcomes, "to try and focus people, to achieve something, deliver it, report it".

The excessive number of OKRs both raised expectations, on the one hand, but was demoralising, on the other hand, when too many of them weren't met by the end of the first quarter under the new DevOps regime.

Bryers at Worldine, meanwhile, reported similar challenges. His IT team were running across a number of territories, and DevOps enabled it to step-up the speed with which it could deliver.

"We accelerated the DevOps/technical side faster than the business could keep up. We focused on the technical people, but it doesn't work unless you've got the business plugged in… I don't think the business was as aware of the pace we were going to do this because they had been used to waterfall," said Bryers.

In other words, the business as a whole was used to the IT team moving at much more measured speeds and was taken by surprise with the changes that DevOps entailed.

In essence, the advice from all three DevOps specialists could be boiled down to: don't try to do too much, don't overwhelm staff with daunting lists of objectives, prioritise objectively and reasonably - and keep the business informed about DevOps and what it can (and can't) deliver.

Missed Computing's DevOps Summit 2017? Stay up-to-date with the latest events at Computing's dedicated events website