“What happens in Vegas, doesn”t stay in Vegas” … a great quote to end the recent DevOps Enterprise Summit 2018 (DOES18) in Las Vegas, urging attendees to take and share the new insights and knowledge gained. This article is an attempt at sharing my discoveries.
This article contains a summary of the challenges that delegates told me about when visiting the GamingWorks stand at the summit. These closely match our global findings, with more than 400 teams having participated in the Phoenix Project business simulation workshop.
This article also describes highlights from the sessions I had time to attend—highlights mostly in relation to the challenges mentioned above, and biased toward toward people, culture and behavior, as this to me is the No. 1 success or fail factor for the successful adoption of DevOps ways of working.
I asked more than 100 people, “What is your top challenge or barrier for your adoption of DevOps ways of working?” I have ordered them based upon the number of times mentioned, highest first. Many of these I captured before the opening of the summit:
- Leaders applying the “pointy-fingered” approach. “You are now self-empowered. Make IT happen.‘
- Lack of “empowerment” and not “walking-the-talk” to foster the right culture.
- Not wanting to let go of a “command and control” management approach and fear of becoming less relevant.
- The need to shift from a “feature” focus to more “outcome” and “business value” focus.
- Product owners “demand” but don”t share why.
- Low focus on “technical debt” and “issues” versus “benefits.”
- Need higher-level product management (business decision makers) to balance “innovation” with “risks” (e.g technical debt).
- Poor end-to-end visibility (including business “before” and “after” activities).
- A narrow focus on Dev and Ops, mostly a Dev “pipeline.”
- No impediment backlog visible for the stream.
- Too much local optimization.
Value and Metrics:
- Too much focus on internal, proxy metrics.
- Need more “value” and “outcome” metrics to gain buy-in to scale up.
End-to-End Communication and Team Collaboration:
- “Them” and “Us” – lack of trust.
- Poor listening and communication skills to explore and understand.
- Lack of end-to-end understanding of the business impact and value we are all working toward.
Poor Management of WIP:
- Poor balance between types of work, feature vs technical debt, improvements, training and knowledge-sharing.
- Engineer disconnect between “what you are working on and the business impact.”
Continual Learning and Improvement:
- Not enough WIP reserved.
- Poor team ownership and ability to structure this.
- No visible backlog of impediments/improvement.
- No prioritization of improvements.
Project to Product:
- Projects claiming resources.
- Product owner resource conflicts and challenges.
- Flip-flopping, context switching work.
Buy-in and Resistance:
- Why should we change?
- Fear, uncertainty, doubt.
- Suspicion of intentions.
- Managers not wanting to change and lose “power.”
Certainly not a lot of CI/CD or technology stuff mentioned here!
What were some of my highlights as they relate to these challenges?
DOES18: Introduction and Opening
In his introduction, Gene Kim revealed one of the goals of the summit was a need to “Elevate focus on spanning business/technology divide,” which clearly relates to a high-scoring challenge. Another topic in the intro was the need to “upskill,” announcing a new skills research project to be launched. The challenges above and the highlights from the sessions give a clear indication of some of the skills needed. Considering the challenges above, I sincerely hope the survey also focuses on business skills as well. In the “Lean Coffee” roundtable sessions, one of the announced topics was “Transformational Leadership,” clearly linking to the top scoring challenges mentioned. More about this topic in my closing.
In a session titled, “Product Management Meets DevOps” a key recommendation was to “build and run teams combined with service ownership,” stressing a need “to connect people to your strategy” and declaring “IT cannot be separate from the business,” and finishing with a need to “manage product value streams, not projects.” (This was the only session in which I heard the word “service”). To me, these highlights clearly make a link back to the challenge teams have in understanding business impact and value. These recommendations will require a changing role for product management to ensure strategic priorities are embedded in the value stream, and a more engaged role of product owners in teams to ensure a shared understanding and focus on “outcomes” and “value” in relation to strategy, rather than on features.
Benefit: Enabling the flow of strategic intent
Mark Schwartz explained that almost all large enterprises are faced with “impediments” that are a barrier to DevOps success, and these are usually people and organizational issues, not technology. He introduced his new “War and Peace and IT” book, and explained an impediment faced by Napoleon in his Russian campaign—Napoleon had “long decision cycles.” By the time he got the right information to make a strategic decision, the fast-changing short decision cycles at the battlefield meant his decision-making was ineffective. The same applies to the way we currently manage IT: long decision-making cycles from idea to business case to program and project resourcing while DevOps teams operate at a fast, agile pace. Mark suggested a need to speed up the “entire lead time from idea to impact” (a topic also explored in Mark Smalley’s latest DASA whitepaper, “The Business Value of DevOps“).
Benefit: Enabling the faster flow of strategic intent
Schwartz went on to suggest three ways to build more agility into the end-to-end value stream and delegate decision-making to a lower level: the “product model,” decentralizing decision to the product team; the “budget model,” decentralizing budget to the team to manage both “innovation” and “keeping the lights on”; and the “objective model,” just give the team the key objectives to be achieved and let them decide how to make it happen anyway they can.
This to me represents a considerable mindset shift from the hierarchical “command and control” management paradigm and requires transformational leadership. In my mind, whatever model is chosen needs to fit with the highlight from the previous session, “Connect people to your strategy.”
Benefit: Enabling faster decision-making around strategic intent
Anders Wallgren‘s session, “Measuring DevOps – The Key Metrics That Matter” was standing room only, with a packed audience taking pictures of all of the slides. I must admit to being surprised, as this is the fifth DOES conference and organizations have had five years to put metrics in place! I can only assume the vast majority were just starting their journey. I was pleased to see not just the traditional metrics often quoted such as “deployment frequency,” “lead time to change,” “mean time to recover (MTTR),” and “change failure rate,” but also business value metrics such as revenue growth; customer value metrics such as customer experience satisfaction; team culture metrics such as education and growth; and the pipeline metrics such as release velocity. This was one of the top-scoring challenges facing delegates who visited our table: “Fhifting from internal metrics to business value metrics.” This focus on business value metrics ties in again to the highlight from the first session above, “Connect people to the strategy,” and the important role product owners have in sharing the strategic intent.
Benefit: Enabling measurement of strategic intent.
Dr. Mik Kersten presented “Project to Product: How Value Steam Networks will transform IT and the Business,” suggesting that project management and traditional cost models are the wrong models, an idea supporting Mark Schwartz’s observations. He went on to add, “Fragmented value streams kill personal productivity,” as well as “impact employee satisfaction.” Silos and proxy metrics not related to business value will also destroy a DevOps transformation. He made some powerful arguments for a complete end-to-end value stream (Not just a DevOps value stream).
These messages were further endorsed by Carmen De Ardo in his session, “Connecting IT to the Business at BMW Group,” during which he also stressed, “Flow time is not just code-to-commit but also, and more importantly, idea-to-value.” Carmen also stressed that the “flow to value” is not just the flow of “features” but also “defects,” “risks” and “debts,” which to me represents a need to manage “value optimization,” a balance of “value creation” and a reduction of “value leakage.”
Benefit: Enabling value optimization of strategic intent.
Damon Edwards, in his session “Operations: The Last Mile,” expanded on the theme of end-to-end value streams, with a powerful case about disruption, context switching, waiting, escalations, lack of trust and employee dissatisfaction—all this because of “four forces that undermine operations” that need to be addressed to empower operations:
- Reduction of Toil—non-valued repetitive work that can be automated.
- Queues, which disintegrate and obfuscate value streams.
- Silos, causing disconnects and mismatches.
- Low trust—the need for “psychological safety, the No. 1 predicate for success.”
Benefit: Enabling the removal of behavioral flow barriers and waste.
Dominca Degrandis in her session, “Making Connections Visible: How to Defrag Your Value Stream,” stressed, “The hardest thing we do is communicate across teams,” making a call to make connections visible as well as “how the different types of work enters the system, where they enter and which tools are used.” This touched on a topic I discussed with DeArdo at itSMF Fusion: the information value stream.
Benefit: Enabling the removal of end-to-end information and behavioral flow barriers and waste.
One of the most powerful sessions for me was the presentation from Dr. Steve Spear: “Discovering Your Way to Greatness: How Finding and Fixing Faults is the Path to Perfection.” He suggested, “Design all work to SEE problems.” Make this part of a philosophy: “See, Solve, Spread (Share knowledge).” He suggested, don’t ask “How is it going?” Ask, “What is not working, and what can we do about it?” This ties in nicely with a finding from the latest State Of DevOps report: “Relentless improvement work leads to excellence and high performance.”
Benefit: Enabling the relentless removal of barriers and waste and the optimization of strategic intent.
DOES18: A Customer Case
Mascha Boender and Stefan Groot presented an ABN AMRO case, “Crossing the DevOps Chasm: From Taboo to Enterprise DevOps,” in which they discussed their bottom-up journey of experimentation. They designed their approaches to see problems. They admitted failure, demonstrating their learning and how they made iterative improvements. They stepped away from a maturity model to a capability model, recognizing that one size doesn’t fit all teams. They provided a “Moonshot workshop,” focusing on specific team goals and ambitions and how these aligned to business strategic goals. They provided a DevOps playbook (based on captured team best practices) to help new teams develop practices at their own pace. To develop team “soft skills” such as communication and collaboration, they used the Phoenix Project business simulation workshop. This was done also with more than 30 leaders to create a mindset shift and demonstrate the change of leadership behaviors required to empower the teams and foster an enterprisewide culture change.
Benefit: Enabling bottom-up and top-down “skills” and “culture” and aligning team goals to strategic intent.
The State Of DevOps
Nicole Forsgren and Jez Humble presented, “The Data Behind DevOps: Becoming a High Performer,” in which they shared findings from the latest “State Of DevOps” report. I won’t go into detail. I suggest you read the report, then read it again. Suffice to say their key messaging echoed many of the other sessions above: “Focus on outcomes, not output,” the need to “identify constraints and continually improve,” and for me an important topic: “Influencing culture through leadership and autonomy.”
The ABN case above was an example of how they gave teams autonomy to change and how they helped change leadership behaviors.
DOES18: Final Thoughts
As I stated at the start, culture to me is the No. 1 success or fail factor. The “State of DevOps” report stressed the importance of leadership on culture. Leadership was the top challenge mentioned by delegates visiting our stand. I”d like to share with you an article on DevOps.com on how to influence leadership behaviors: The article is written through the lens of “communication,” tying back to a comment from Dominica Degrandis: “The hardest thing we do is communicate across teams,” and tying in with the ABN AMRO case about developing communication and collaboration skills.
And finally, getting back to the start of the summit and the need to “upskill,” with the announcement of a skills survey, I’d like to provide some input. Here is a DevOps Agile Skills Association (DASA) article, “DASA Provides Core Skills Key for DevOps Success,” which summarizes the “people and culture” elements from the “State of DevOps” report (this year and previous years) matching these with findings from more than 400 teams having participated in the Phoenix Project DevOps simulation. This I believe can provide input into the survey.