Happy New Year one and all! Whilst time, or at least the Gregorian calendar, may be a human construct, I do enjoy the annual opportunity for navel gazing and setting intentions for a new dozen of months. I’ll be working with Zen koans daily in 2019 as part of my evolving interests in meditation and the brain. This however, represents my thoughts about the world of DevOps in the coming year, and is also a reflection on my thoughts on the same a year ago.
The key DevOps topics I think we at Ranger4 will be tackling in 2019 are:
- The demise of the DevOps team
- The rise of the DevOps Dashboard
- The DevOps Target Operating Model
- Extension across the enterprise & leadership
- Value Stream Mapping and Flow
- Value Realisation
- Ending project based funding
- Reversing outsourcing and avoiding a skills war
- DevSecOps and software liabilities
Let’s have a look at each of them in some more detail:
- The Demise of the DevOps team
In my 2018 predictions (check out number 7), I said that we’d see a dip in the number of DevOps teams reported. Whilst there wasn’t a decrease, for the first time since The State of DevOps annual surveys and reports began, we didn’t see an increase but a flat 27% year on year. We think this is good news as an indicator that the message is spreading that DevOps is everyone’s job.
2. The Rise of the DevOps Dashboard
We’ve been thinking about DevOps Dashboards for a long time at Ranger4 and helping our clients build them according to the measurements their current fluency indicates are best to learn from using the systems they already have like Azure DevOps and Grafana and some have adopted Hygieia for this purpose. The outcome we are looking for is enterprise wide capability to measure to improve value realisations and the enablers are visible, accurate and real-time metrics. It’s tough to do though with the multitude of tools available and legacy systems not following the ‘telemetry everywhere’ principle. From the back end of 2017 though, we saw vendors such as XebiaLabs and ElectricCloud starting to use their tools as a kind of data fulcrum that we believe will ultimately make it easier for people to curate, share and act on the intelligence their DevOps toolchains are generating. The constraint is that organisations typically won’t access these dashboards if they are not already using the vendors’ products. We’d love to see a truly independent, accessible and configurable DevOps dashboard in 2019 that makes it quick and easy for teams and global organisations to measure their improvements. Please do let us know in the comments section if you believe you have one.
3. The DevOps Target Operating Model
As organisations adopt DevOps practices and seek to break down silos and dependencies, there is a realisation that the traditional hierarchies and organisational structures present a constraint. As we work to distribute authority and drive autonomy and empowerment, it becomes increasingly apparent that layers of reporting lines and management are a command and control constraint. Whilst we move towards product centric, autonomous team models the challenge posed is how to get from where we are now, to where we want to be.
Last year I mentioned AI, and called it a red herring. I’ll stick with that as we haven’t seen any of the organisations we work with reach a level of fluency where they are using AI on a day to day basis. 2018 did see, particularly towards the latter part of the year, an increase in noise around AIOps. The noise is coming from vendors like Moogsoft, StackState, OpsRamp and Splunk. A Gartner term defined by BMC as: “multi-layered technology platforms that automate and enhance IT operations by 1) using analytics and machine learning to analyze big data collected from various IT operations tools and devices, in order to 2) automatically spot and react to issues in real time.”, I think we could see organizations with higher fluency picking up these tools by the back end of 2019, particularly if their existing vendor partners, like Dynatrace, make it easy for them to access and consume.
What concerns me most about this trend though, in the context of DevOps, is that it separates Dev and Ops (although last year I said we would see more ops in DevOps - see number 4) and in my mind this capability should be part of a DevOps Dashboard.
5. Extension Across the Enterprise & Leadership
This theme has come up everywhere all the time in 2018. The core challenge is that agile, DevOps and digital ‘transformations’ (we really prefer the word ‘evolution’) are being driven by technology teams in order to better serve the business (and ultimately the customer). The problem with this is that the business oftene doesn’t know why we are doing it or sometimes that we are doing it at all. This means that when we ask them to provide product owners or work with us on a product backlog we see scenarios where product owners ‘have a day job’, requirements are still sent through in large batches with milestones and deadlines and project based funding (see 8). The improvement is what I called BizIT in last year’s predictions.
Exacerbating this is a conflict at the leadership level. Whilst teams can be commanded to be empowered (an oxymoron, I know), they need to be helped to unlearn the behaviours they have practiced for many years. This requires a certain kind of enlightened leader and our experience with leaders is that many consider themselves superior to learning and find it hard to coach - their ingrained behaviour is to tell others what to do. Since we are mainly reliant on these leaders to take the DevOps message organisation wide, we need to tool them up too.
6. Value Stream Mapping and Flow
Linked to 5 is this - helping organisations and individuals to collaborate together to visualise the flow of work through their value stream, and by doing so identify constraints, waste and bottlenecks, is a foundation stone to engaging across the whole enterprise. This does require socialisation of the idea that DevOps is concerned with the end to end value stream - not simply starting at the point at which an idea hits a product backlog and moves into development - in order to get the requisite people into the room. It also requires time commitment from all participants into the process and recognition going into the facilitated exercise that this is the start of an ongoing improvement cycle where all will take accountability for the changes agreed as a group - not a one off event that people participate in for a couple of days and then wander off and forget all about.
7. Value Realisation
Linking to 6 then is this - the focus on the realisation of the value in the customer’s hands and the fast flow of that value. Organisations with lower DevOps fluency continue to be focused on uptime and number of defects as the key measures of their success to their ‘business’; by having all working together on a single value stream allows us to refocus efforts on value flow cycle times and make more meaningful improvements and decisions. This brings us right back around to DevOps dashboards as what we want them to show is the business value of the user stories delivered, compared to their projected value. This closes the feedback loop and gives us a view on end to end flow and enables humans to make informed, evidence based, decisions about what to do next.
8. Ending Project Based Funding
Okay, so I’m probably being way too ambitious and optimistic here for this to happen in 2019, but seriously people, we gotta stop. When we think about the DevOps target operating model, broadly described as cross-functional, autonomous and product centric teams, to have them effectively work in small increments off a product backlog means that requirements need to arrive in small, frequent batches (just like how we want to deliver them) - not large batches driven by a project mindset. This means that we need our funding models to align too - they need to be continuous and capacity based. This is a major mindshift that requires true end to end value stream coordination, collaboration and trust. I’m reading Mik Kersten’s book ‘From Project to Product’ at the moment which uses case stories from Nokia (bad) and BMW (good) to show how and why this is essential as we all go through this period of evolution - it’s my recommended read for 2019. Along with ‘Where the Crawdads Sing’ from Delia Owens if you want something completely different and unconnected to the technology industry.
9. Reversing Outsourcing and Avoiding a Skills War
The 2018 Accelerate State of DevOps Report uncovered a negative correlation between outsourcing and organisational performance - one we have observed for quite some time, but, as ever, it’s extremely useful to have the data as evidence. Whilst we understand that, as with the siloing of IT organisations in order to manage cost centres, that outsourcing presented a logical choice for managing the technology spend in an organisation, the movement now to a digitised world and the recognition that technology is not a cost centre but the strategic enabler, demands that we strategically invest in engineering.
There are many problems with outsourcing in a DevOps world, mainly around contractual commitments, bureaucratic processes, colocation, cultural and linguistic translation and longevity of people in roles that all lead us to a point where many major enterprises are openly reversing this trend and planning to reinvest in their own, internal engineering expertise. Whilst we support this as the preferred direction of travel, there are also a finite number of resources available so we could see the start of a war over skills in local regions where skills are in high demand. We have already noted this on the ground, particularly in some of the large cities in the north of England. Perhaps there is room for a new model where organisations take their DevOps target operating models to some of the lower cost geographical locations and invest in building local teams and cells and collaborative technology solutions on a global basis.
10. DevSecOps and Software Liabilities
The subject of DevSecOps became very much alive in 2018; in the UK we saw the inaugural DevSecOpsDays event in London and we saw an increase in our customers getting infosec, security and dev and ops in a room and an uptick in all levels of organisations adopting DevSecOps automation. We are hoping to see this trend continue through 2019; security has felt like the last bastion from both a cultural and automation perspective so getting these guys on board represents some serious constraint smashing. Meanwhile, the conversation around software liability keeps on getting louder pushing DevSecOps from afterthought to legal imperative.
So there we have it. I’ll put my crystal ball away for now and get on with the doing. See you back here next year. Thanks for reading and I hope you have a DevOpstastic 2019.