The State of DevOps
It takes a lot of work, change is hard, progress is slow, but the results are worth it.
Join the DZone community and get the full member experience.Join For Free
James Governor, analyst and co-founder of Red Monk (@monkchips) kicked off the event by providing a 2019 Market View on the development velocity for organizations entitled, "How to Be Elite versus Struggling." CI/CD is the key to moving faster. Agile is part of organizational transformation – smaller teams, faster iteration. DevOps is creating an environment in which development and operations work more closely together. All of this is happening in pockets – some parts of the organization are moving quickly but not the whole organization.
Containers and Kubernetes (K8s) are enabling continuous delivery. CNCF is driving the cloud-native landscape. Hybrid has become a thing with OpenShift, Anthos, Pivotal providing hybrid K8s-based infrastructures. GitOps is now a core technology, all young developers grew up with Git, automation from the perspective of the work developers are doing, part of the primary transformation of how we become more effective.
Observability is improving with more sophisticated monitoring enabling distributed tracing of the platform and the development process. Organizations want to know how they are performing, whether or not they have improved? Are we delivering more applications with better quality? Every company should be data-driven – if you have telemetry, you should take advantage of it to move to progressive delivery from continuous delivery (CD).
Anders Wallgren, Chief Technology Strategist, CloudBees talked about how “Release” is the enterprise challenge. With one team and one app, a release is not the challenge. However, multiple teams working in independent and interdependent systems and applications, overlapping stacks, conflicting schedules is extremely challenging. Only 50% of organizations have gotten to the point where they’re even doing DevOps/CD. Only 2% to 6% have fully-automated testing. You cannot achieve velocity with manual testing. Organizations are using continuous integration (CI) more than CD – it’s a lot more work to get to CD, it involves more people and more steps. CxO’s are vastly overestimating the maturity and adoption of Agile, DevOps, and the best practices inherent in them. With the DORA statistics showing the accomplishments of high performing organizations – there is tremendous motivation to implement a DevOps methodology though it's far easier said than done. Focus more on automation, especially test automation, to put more value at the customer's fingertips more quickly.
Stas Zvinyatskovsky, Managing Director froom Accenture, notes that DevOps enables organizations to achieve speed, quality, productivity so they can compete. Companies are facing continuous disruption from all angles. They need to develop fast with quality and high productivity. Adopt a set of values and practices to enable continuous delivery of flow — maximize the volume flow from concept to cash. Automation and decentralization are the two hammers with which every company will do this. Decentralize architecture and automate to maximize the flow of value.
Daniel Ritchie, Distinguished Engineer at Broadridge, believes the problems you are having with teams can be mitigated with DevOps. Broadridge formed a DevOps center of excellence to change the culture and promote collaboration as they scaled their DevOps initiative.
George Swan, Senior Director of Build Platform of Autodesk, has been in charge of CI/CD for the last four years. Autodesk is now a cloud platform company with 10,000 people, 4,000 in engineering, providing automation of CI/CD for software, documentation, and APIs for architecture/engineering/construction (AEC) firms. It can be deployed to the desktop, cloud, and mobile and the CI/CD environment is common across all products. The cloud product is based on containers. The platform is composed of GitHub, Artifactory, and CloudBees/Jenkins for CI/CD. Over time, George expects they will migrate away from the data center to AWS, providing pipeline-as-a-service, and building out a data store of data and providing insights back to developers.
Rajeev Mahajan, CIO Engineering DevOps, HSBC runs DevOps as a function and strives to be an invisible team as they work to put the power in the hands of the developers rather than separate functions. Rajeev and his team currently manage four stacks: mobile, distributed, mainframe, and mid-range. If all four are not equally Agile, they are not able to run seamlessly. The goal is happy, smiling developers – it’s more about the cultural transformation. We want to be a great place to code. We want to build digital experiences internally – build great experiences for customers. Transformed developers can to amazing things to transform the business. Currently running 90 tools, and anticipates reducing that to five or six with a platform.
Swati Shah, SVP of Emerging Technology at U.S. Bank wants to improve both the customer (CX) and developer experience. To do this, she focuses on three things: 1) enterprise human side; 2) enterprise technology; and, 3) cost. The enterprise human side is about feelings, behavior, collaboration, and the definition of shared goals. Enterprise technology is about platform thinking, defining what we want from DevOps, how the customer will be impacted. the developer experience, onboarding, transparency, tools, and processes. Costs involve soft and hard benefits, what's necessary for the early adoption of CD, and the soft benefits of using open source to change the benefits of the engineering organization.
Adam Robertson, Head of DevOps of Pinger has 80 committers, 140 employees, and a lot of heritage code. Adam believes the successful adoption of a DevOps methodology is a people issue, not a technology issue. The key is getting people on board with change.
Shawn Ahmed, V.P. of Product Marketing at CloudBees emceed a panel discussion among the customers:
What is the most misunderstood part of DevOps?
It doesn’t have anything to do with people because ultimately it’s a culture and mindset and everything is automated with no humans involved.
Breaking down silos.
Given the complexity of different systems, we have to get along to deliver a meaningful product to the customer
It’s different for everyone.
It’s automation, get technology out of the way to write good code and produce a good product.
- Make sure engineering organizations perform at their best.
What are the outcomes for DevOps?
- Start small and scale fast
- Number of releases
- Number of incidents
- Dev experience, CX
- Use design thinking to ensure outcomes and outputs are what’s desired
- People have vague ideas of where DevOps is going to take them – we have four goals:
- Employee engagement
- The metrics are easy to collect if you put your mind to it
- Output metrics are important but they need to be evaluated in context
- Lead time to deploy
- Meantime to deploy
- How fast can I get it into the hands of the customer
- How do developers like doing their job, happy and smiling developers, remove blockers and waste to start doing what they want to do as early as possible
Can culture be created in advance?
- You need the tools and capabilities in place before you get to culture.
- You have to invest a lot with yourself and your leadership team to change the culture
Where do you see DevOps going in the next six to 12 months?
- I don't see revolutionary changes
- More focus on the ops part
- Expand beyond the commit to the production pipeline
- With K8s, the law of preservation of complexity playing out
- The same is happening in the delivery pipeline
- Delivery pipelines will become invisible and cloud providers will obfuscate the pipelines
- The barrier to entry getting into cloud-native is getting lower and lower
- Most of the work heretofore has been focused on the dev side, a lot of work is needed and will be done on the Ops side
- Baking in security
- CD will mature
- Cloud pipelines, hybrid multi-cloud
- Chaos pipelines to create chaos and see how resilient your pipelines are
- A shift toward and focus on ops – infrastructure as code, the maturity that developers have does not exist on the ops side
- Abstraction of complexity
- Leveraging data for self-healing systems
- Maturity around AI and modeling
What does the future of ops look like?
- Full automation
- Alerting, monitoring, storage can all be managed as a code
- We won’t see a major shift in the role
- The DBA team will be responsible for the deployment of DBs shifting from manual deployments to offering a service
- The world of manual QA and test is a good example of what will happen to ops
How do you measure the ROI of DevOps?
- We measure in terms of developer happiness
- Speed, quality, productivity, employee engagement
- Establish the metrics business cares about
- An investment is required, are we able to do more with less to produce better quality
- DevOps goes beyond traditional ROI
- Start small, have confidence from engineering, C-level, and teams that we’re making investments in the right direction
- You need a solid platform that can scale fast
- Need to define the strength and culture of your organization
- Contribution, communication, and collaboration
- Some teams do not need to deploy multiple times a day
- A cloud platform helps to achieve it
- A web dev team doesn’t see the need to deploy more frequently than once a month – customers don’t want updates more frequently than that
- Help people feel like they’re part of the decision-making process
- Provide a solution that’s so compelling that people see it’s obvious it needs to be done
What is your company’s DevOps superpower?
- People - How fast we want to learn, build a new generation of technology
- People with experience helping others transform
- The ability to be humble, take a step back, be willing to consider alternatives, listen, everyone has a voice, and not dictate what happens
- We're an engineering-driven company and good ideas float to the top
- We have the ability to take ideas and act on them, able to make change happen, follow-through, execute and innovate
- We do not ignore the small ideas, small things accumulate into a superpower, super enablement, super happy developers
- Cross-functional integration, working with a new team every three days, able to bring together different groups, integrating products as well, able to digest different technologies and ingest
How do you handle governance, risk, and compliance?
- De-risking what we are doing
- What we don’t want to do – send crappy code out faster
- Full trustability from the first line of code for engineers, compliance, and risk
- Build a more compliant pipeline
- A lot of existing procedures require manual intervention, can take a year to change or not achieve real continuous delivery
- Auditors and compliance are ten years behind DevOps
- Build SOC2 certification into the delivery pipeline
- We spent a year upfront having a discussion with stakeholders to define a process that gets you 90% there
Opinions expressed by DZone contributors are their own.