Principles and Practices of DevOps Track

Andrew Clay Shafer
Track Chair

What changed when application lifecycle transitioned from a software artifact shipped in shrink-wrap to a highly scaled, highly available service oriented world? DevOps is the label for the professional and cultural movement as well as the principles and practices of the high performing teams who are literally setting new standards in what is practical and possible. This track will shed light on the tools, processes and cultures that are evolving on the edge of the DevOps movement, and how that is impacting what is happening in both startups and the enterprise. Developers, system administrators, testers, product owners, and technical managers will gain an understanding of what is possible with the best moderns tools and processes with perspectives, principles and practices to optimize your organizational workflows.

The DevOps track will explore topics like:

collaboration between developers, operations and other contributors to the value stream
automation of provisioning, configuration and application deployment
monitoring and metrics driven feedback loops
balancing risk strategies (minimizing incidents vs minimizing the mean time to recover)
fully automated development, test and production pipelines
deploying and operating micro services
cross functional information radiators and cultures of sharing

Breakout Sessions

Steve Speicher OpenShift Red Hat John Wiegand Senior Technical Staff IBM Rational		Continuous Delivery Pipeline for Hybrid Cloud Tuesday: 9:45 - 10:30am For development teams to meet the pace of business needs, tools are cobbled together with focus on just "getting the job done". This works well, for the specified scenario and for a limited period. Short cuts on approach, leveraging private APIs or emerging open source packages. In this session, we'll take a look at a few examples and what went wrong and why. We'll outline a simple approach to solving these problems, that don't take away from the qualities that led the developers to "get the job done" but to be able to get it done better, with less rework. The approach outlined is really about using what works and focusing on: code speaks. Working in the open, on source code, to prove out tool integration scenarios...starting with a very simple and modest start. We'll highlight the importance of keeping it simple but a way to layer on complexity without impacting the core value and simplicity for tool APIs. It will go from starting with basic pipeline-to-tool integration, to how tools can hook into the pipeline while still providing traceability across the different artifacts produced or modified from different stages, as well as other scenarios. We'll highlight how people can leverage and contribute to this effort as well. There will be a couple of demos to highlight some of the integrations.
Dominica Degrandis Director of Training and Coaching LeanKit		Staying Focused with Conflicting Priorities Tuesday: 10:45 - 11:30am When workers ask about priorities, a “do all the things” answer can do more harm than good. Attempting to increase the workload through a pipeline that can only handle so much is a futile exercise—it only increases interruptions that cause further delays. Clear priorities are vital for reducing expensive context switching. But they are tough to get when competing requests from many customers arrive simultaneously. This talk addresses what clear priorities can do for teams and how a sensible prioritization method helps teams increase throughput through a pipeline.
Chris Brown Gun for Hire		What is a "DevOps Transformation?" Tuesday: 1:45 - 2:30pm Over the last several years as a software leader developing Amazon EC2, and as a CTO of a "transformational business", I've heard the same stories from an increasing number of F500 companies and enterprise customers. They use words like "journey", "transformation" and "aspirational". What are they doing differently today, or at least hoping to do? They are redefining their businesses based on code and operations, leveraging large-scale IT to rapidly delivery new services and applications to market. DevOps is not about tools or practice. It is a psychological reaction to this shift. Over the coming years, becoming a "coded business" will be the focus not just of the web elite, but of most major enterprises. I would like to explore this story with you, discussing the business, technology and people who are a part of this change.
Allan Wagner Technical Marketing Manager IBM Rational		Continuous Testing in Today’s DevOps World Tuesday: 3:45 - 4:30pm DevOps is often described as extending agile and lean principles across the entire delivery lifecycle to continuously release increments of higher quality software at a lower cost and with less risk while delivering a positive user experience. However, in this description and in the new DevOps world, testers often seem to struggle with fitting in - being considered an equal. This is terribly unfortunate when you consider that testing makes up at least 1/3 of the delivery lifecycle and when that Sev1 defect does slip into production it is the tester who is "called out" wearing the “badge of shame”. Even though, they didn’t write the code, they didn’t write the bug, and they aren’t responsible to fix the bug. Join Al as he puts the Test in Dev+Test+Ops sharing techniques and the tools available which can help organizations solve some of their testing challenges and work to achieve continuous testing. Hear how to bring testers into the development process earlier, build quality in, drive out defects sooner, and increase test coverage. Combined with deployment automation, leave with an understanding of how continuous testing is the enabler to continuous delivery.
Andy Troutman Development Manager Amazon		Software Delivery Patterns in the Cloud Tuesday: 4:45 - 5:30pm Immutable infrastructure, blue/green deployments, configuration management, rolling updates, AMIs, containers… The transition of workloads to on-demand capacity in the cloud has spawned a growing set of options for delivering software updates. In this talk, we’ll identify some classic and emerging patterns for managing and releasing new software in the AWS cloud. We’ll take a look at tradeoffs between deployment safety, speed of delivery, and maintenance effort, to try to identify an approach with the right balance for your team. Throughout the talk, we’ll look at case studies for implementing the various approaches using open source tools like Chef and AWS services like CodeDeploy.
Mike McGarr Engineering Manager Netflix		Freedom and Responsibility at Netflix: Centralized Team in a Decentralized World Wednesday: 9:45 - 10:30am Netflix is famous for its culture of Freedom and Responsibility, which empowers engineers to choose the best tool for the job. Engineering teams at Netflix can follow the latest industry trends and sometimes they even create them. No one single team is responsible for innovation: Does this generate opportunity or waste or both? The Engineering Tools team helps to direct other teams’ experimentation toward new products while preserving their own ability to innovate. In this talk, Mike will explore the challenges of building and operating a centralized service team in the decentralized microservice world of Netflix. This includes building tools customers love, standing up an effective customer service model, productizing innovations, enabling self-service tooling, and providing consulting services to engineering teams. Mike will share tips on how to effectively establish a “tools team” in your organization. Related Resources: GeekWire Interview: Deploying code The Netflix Way Session: Operational Insight: Concepts and Real-World Applications
Pete Cheslock Sr Director, Ops and Support Threat Stack		Continuous Security Wednesday: 10:45 - 11:30am The same junior level operations engineer who can accidentally provision thousands of systems and blow your budget away, is the same person who can make one small change to a security group which now allows all access to your backend systems. The long contested battlefield of remote access to production machines has only gotten uglier since the rise of The Cloud, which has obliterated the line between building the system and running the system. “Lock out the developers” is not an acceptable policy anymore. Developers inherently build better systems when they experience running them. If the goal is frictionless releases, then we must build a trust but verify model of access control. The wall that existed between Engineering and Operations has largely been broken down with both sides embracing change. But now we’re faced with a new wall, one that exists between Security and the rest of the technical organization. Fortunately, there are collaborative ways to overcome this wall. We'll discuss the role of security in this new *aaS landscape. We'll talk about things to do when you have a dedicated InfoSec team, and tools you can use when you don't. We'll explore what it means to build in security in the same way you build in quality as part of your continuous delivery pipelines. And how you can strengthen your security posture while maintaining your ability to move quickly and deliver value to your customers.
Dan Piessens Sr. Agile Consultant Centare		From Rags to Riches: Transforming a Team with Continuous Delivery Wednesday: 1:45 - 2:30pm So you’ve setup a CI server and automated deployment. You’re sitting back admiring the builds flowing out, but your team looks at you like they’re in the pilot seat of a commercial jet and you’ve asked them to fly it! All this new feedback is streaming back to them, and concepts like branching, large databases, and releasing the whole application stack at once are becoming a challenge. To top it off the team wants to increase quality, but they don’t know when to fail the build and want to change the quality gates whenever they are time pressured. I’ll cover what feedback is valuable and dysfunctions that develop both in the team and the operations group as a result of continuous delivery. Then we’ll discuss how your team can capitalize on the information to improve themselves and discuss how the applications can evolve to replace concepts done in traditional delivery today.
Arnoud Lems Agile Delivery Lead Avanade		Automating Deployments with Desired State Configuration Wednesday: 3:45 - 4:30pm Part of the devops mindset is the notion that “configuration = code”. Meaning you should treat the configuration of environments the same way as you would treat Java or C# code. This session will show you how to use desired state configuration to automate not only deployment of applications but also to automate server configuration usually done manually by the operations team (e.g. deployment and configuration of a webserver). Based on a scenario where we look at how a .Net REST service is developed, deployed to a test environment and ultimately go live on an Ubuntu server on Azure the session will explore how to add a devops mindset to the ALM domain.
Matt Stratton Solutions Architect Chef		The Five Love Languages of DevOps Wednesday: 4:45 - 5:30pm The title of this session is inspired by the self-help book “The Five Love Languages” by Gary Chapman, but the similarities end there. When we are working to bring about cultural change in our organization, it’s essential for us to understand that not everyone speaks the same “language” of DevOps as we do. The CFO has different drivers and communication style than your friend in the QA department, who is still different than the DBA in that remote office. In this talk, I will work to help provide techniques and strategies for identifying the drivers of your peers and decision-makers. We will talk about beliefs, perceptions, and communication styles, all with the express intent of helping YOU drive home the value of the DevOps philosophies and work-practices to make your organization succeed!
Sam Fell Product Manager Electric Cloud		Making CD Safe for Ops: Introducing Monitoring Earlier in the Agile Cycle Thursday: 9:45 - 10:30am Are your developers Agile, creating more frequent releases? Are your Ops teams nervous, worried about performance issues being introduced into production? Join CTOs from Electric Cloud and Boundary as we discuss best-practice patterns and processes Dev & Ops teams can embrace to: Gain predictability and reliability by standardizing deployment processes across Dev, Test and Production deployments Eliminate performance problems and improve quality by monitoring applications earlier in the cycle Empower Ops teams and give them more confidence in application updates by providing shared control and visibility of the end-to-end deployment pipeline
Roy Rapoport Manager, Insight Engineering Netflix		Operational Insight: Concepts and Real-World Applications Thursday: 10:45 - 11:30am What are monitoring systems for? All too often, we deploy systems which perform relatively dumb functions -- take data in, spit it out (sometimes in the form of graphs), and harass us in the middle of the night when some arbitrary metric goes below or above some arbitrary, carefully-selected, threshold. In this talk, we propose a conceptual framework to describe the utility of monitoring systems, covering value creation at the platform, visualization, and analytics levels and tying them to the concerns of their users. Once such a framework is described, we intend to demonstrate one real-world manifestation of these ideas -- Real-Time Operational Insight systems at Netflix, where telemetry is a 1.2-billion-metrics-per-minute business, and Insight is considered a first-level, business-critical function. While some of the engineering described requires significant specialized knowledge and investment, our goal is to provide relevant and meaningful information usable for anyone working in an environment where applications are operating in production environments.
J Paul Reed Consultant		The Changing Role of Release Engineering in a DevOps World Thursday: 1:30 - 2:15pm There is no denying that DevOps has shaken up the world of developing, deploying, and operating software. But with all the buzz around new techniques and technologies, it's easy to get lost in the "We deploy hundreds of times a day!"-cacophony and the new "DevOps tools." In this talk, we'll look at the conversations DevOps is revitalizing around age-old topic in release engineering and application life cycle management, plus look at the aspects of software delivery which DevOps does not magically solve. Whether you're a release engineer wondering how you fit into "the DevOps" or you're responsible for the release engineering function in your organization, we'll explore what the new world looks like for this critical role... and which aspects of it our industry is leaving behind.