Topconf’s second annual Baltic DevOps conference in Tallinn in May attracted 120 attendees who could choose between talks of a technical nature (which attracted 80% of the attendees), and talks about how to organize DevOps. As a member of the programme committee, I had advised about the content of the organizational track. We had compiled an interesting mix of speakers and presentations, and this is a summary of some of the points that resonated with me.
Before the attendees got to choose between talks in the technical track and the organizational track, they listened to the opening keynote that was about the growth of a start-up in the Fintech sector, TransferWise, represented by Alvar Lumberg and Edgar Maloverjan. As is often the case, this organization was trying to find a balance between giving the workers as much autonomy as possible, while addressing overarching ‘enterprise’ needs, such as compliance to industry regulations. They have organized the work into several small vertical product-oriented teams, and teams that provide the platforms that the product teams use. One of their guiding principles is to optimize for reducing the mean time to repair/recovery, rather that increasing the mean time between failures. In other words – even in the Fintech sector – they are not afraid to fail (within limits, of course).
‘You build it, you run it’
Eduards Sizovs delivered the first talk in the organizational track and spoke about 8 things that make continuous delivery go nuts. Just as mentioned in the keynote, he regarded the concept of ‘you build it, you run it’ as a game-changer. His take of organization was similar to the TransferWise case: cross-functional, product-oriented teams, supported by infrastructure-oriented teams. He encouraged a pragmatic adoption of industry practices, always keeping the focus on optimizing the flow of value though the whole organization. A final take-away that I’d like to share is the importance of understanding the customer perspective, which can be achieved by external customer visits but also by visiting the internal customer service department.
Yegor Bugayenko spoke about negative side effects of rigorous implementation of continuous integration, and how to pre-empt some of the problems by using a ‘pre-flight build’ approach. Without going into the details of his talk, I realized that continuous integration is very much work-in-progress and that the priority is to find a workable solution, rather than the fully-automated utopia.
Fear of release
Antons Kranga has been involved in DevOps before the term DevOps was coined by Patrick Debois in 2009. He has more insight than most so-called evangelists and presented DevOps with smell. It referred to sensing or ‘smelling’ unproductive anti-patterns in organizations and sharing his thoughts about how to address them. He was very conscious of the human condition, for instance the ‘fear of release’ that most people have, which can be a bottleneck in trying to improve the flow of work from development to operations.
I also had a slot, and facilitated a workshop in which the attendees discussed the kind of behaviour that development would like to see from operations, and vice versa. The results are summarized here but the major points focussed on fostering a collaborative culture with transparent sharing of information. We briefly discussed which drivers of behaviour would need to be changed and identified three. Providing more background information helps people to understand how their work affects others, and therefore helps them to make the right choices. Working in the same physical location is usually beneficial. Finally, multidisciplinary collaboration is improved by walking the other party through your thought-patterns so that they understand the ‘why’ that drives your behaviour. Some thought-patterns also need to be revisited – we often fly on automatic pilots that worked well in the past, but are not necessarily effective in our current environment.
Jan de Vries spoke about how to let the business discover that they need DevOps, which is a good way convincing IT management that DevOps is a worthwhile investment. Jan also touched briefly on the topic of anti-fragility, referring to the excellent Sogeti trend report Design to Disrupt. Anti-fragility is a fascinating topic contrasts robustness and fragility and that can be summarized by Jez Humble as “Fragile wants tranquillity; Antifragile grows from disorder; Robust doesn’t care too much”.
Andrey Adamovich delivered the final presentation, an enterprise flight into DevOps space, that contained some good tips for achieving more effective collaboration between development and operations, for instance that development should contribute more to operations topics such as logging, configuration and monitoring, and the operations should improve their analyses of monitoring data, providing information that is valuable for further development.
Latest posts by Mark Smalley (see all)
- DevOps in a Baltic business context - October 2, 2017
- IT-Wise, Business-Foolish? Time to Look at the Bigger Picture - May 22, 2017
- A Journey from IT industry Guidance to Business Value - April 11, 2017