Armory Platform Success Playbook: Stages of Software Delivery Evolution Infographic

Apr 7, 2017 by Ben Mappen

Before you begin: Read our Stages of Software Delivery Evolution Infographic to understand which stage your company currently falls into.

Seven Steps to Success:

  1. Treat deployments as a product. Assign a product owner (ideally full-time) who is responsible for ensuring the success of Spinnaker within your organization. Just like other product managers in your company, this person will talk to internal customers, define requirements, build a project plan, articulate a product road-map, and ensure your internal customers’ success.
  2. Achieve consensus that a deployment problem exists (1-2 weeks). You probably have a good idea of what deployment problems exist but it’s important to get consensus from the engineers on those problems, in their own words. We strongly recommend surveying your service owners and engineering managers to help determine the pain points that everyone is experiencing. It’s hard to solve a problem that no one perceives to exist. Additionally it’s good to illustrate the pain points that applications teams are experiencing today as a way to motivate teams to move to Spinnaker.

Questions for engineering leadership:

  • What are your strategic engineering initiatives for FY18/19? Which ones are in jeopardy?
  • How much revenue / broken customer trust have failed deployments cost you in FY17/18?
  • How is your success measured?
  • How are you measuring your teams’ velocity today?
  • Which of your teams are deploying the most right now?
  • Which team is deploying the fastest? Which has the most failed deployments? The most manual judgements? Which one has pull requests open the longest?
  • How are you measuring application health today? Do you have SLAs or SLOs for all your applications? Even internal apps? Do you get alerted when application health is impacted? Can you easily roll back to restore health?
  • Are you able to canary today? Is it fully automated? Do humans have to look at dashboards for hours or days?
  • How many deployments do you do per week? Per month? How many do you want to be doing?
  • How much time are you spending on manual judgements?
  • How much time/resources have you spent on implementing/maintaining CI/CD? How about moving forward?
  • Do you have a full, comprehensive view of your entire SDLC today? Where does your SDLC break down? Where are the gaps an inefficiencies?

Questions for application teams:
We recommend using this survey for your application teams, but you may choose to use your own.
Here are some questions you might ask your application developers:

  • What service do you currently work on?
  • How do you deploy today?
  • What pain points exist with your deployments?
  • How urgent is it for you to solve this pain point?
  • Are you interested in participating in the first install?
  • What do you value more? (velocity | safety)After you complete the survey we will help you compile the results and disseminate it across the organization as a solid starting point.
  1. Define success metrics (1 day). Good candidates for success metrics to track are: number of deployments, time to deploy, reduction of manual steps, or improved SLA. Choose the two or three metrics most germane to your organization based on what you learn from surveying application teams and highlight those metrics as you roll out Spinnaker in production.
  2. Define and execute on a tight proof-of-concept (1-2 weeks). Unlike other developer tools like Github or Pagerduty that can be seen as a more “personal choice”, Spinnaker represents an agreement between an entire organization on how you will operate. Spinnaker enables the DevOps vision of empowering engineers to truly own their applications and deployments.
    For some companies this will be a sweeping cultural and philosophical change from how you’ve traditionally operated. From our experience, the best way to prove that this is a better way of working is with data (see step #3 above), and the best way to quickly acquire data is by executing on a first install that demonstrates success (ideally this first install takes 1-2 weeks at most).
    We recommend starting with a single team and a single app being deployed to production with Spinnaker. We’ve found through working with other enterprise companies that it’s critical for the application to go all the way to production (and not just to a staging environment) to help prove that this works and that the data is real.
  3. Choosing a right team for the initial install. You probably have a good idea of what teams might make good candidates for your first install. Ideally, you find an application team that has an existing and urgent pain point with their deployment system and who is motivated and incentivized to improve it. This team will become your early evangelists and help spread the word about Spinnaker and why it’s valuable.
  4. Choosing the right application for the first install. There are some applications better suited for Spinnaker than others. We will do an additional survey that asks application teams about their current deployment practice(s). Some questions would include:
  • What external data sources (databases, API, datastores) are needed to run your app?
  • How is configuration management handled for your app?
  • How are data migrations (if any) managed for your app?
  • How is service discovery handled within your app?
  • Describe the build process and its outputs?
  • What type of automated testing is built into the app?
  • How is the system configured?

Ideally we find an application that has few external data sources, simple and smaller configuration management, no data migrations, ELB based service discovery, and a clean build process which has simple output like a WAR file (Web application ARchive) which would make a great first install candidate.

  1. Build a case study (1 week). When the first install is completed, compile the results and present them to executives for additional buy-in and support, and also to the next set of application teams to identify teams for the beta launch.
  2. Beta launch (1 month). The beta launch will open up Spinnaker on-boarding to a larger group of teams. If the first install team can be seen as the center of the circle then the beta is the next concentric circle outward; as such, it will include additional features and functionality. Since you already surveyed the application teams in Step #2, you know who you should approach first for participation in the beta.
    Hopefully, by now your first install has generated enough buzz that beta customers are banging on your door to get in, but if not you can use your case study from Step 5 to help sell them.. In order to execute on the beta launch successfully, you’ll need to create a more robust project plan and build the functionality needed to support the larger set of customers. Set a hard deadline to launch the beta and rally your platform team around this date to make sure you hit it. Your product owner (Step 1) will be largely responsible for making sure you hit this date, on-board, and train the new teams successfully.
  3. General availability (1-3 months). After the beta launch is completed, you’ll have even more data and case studies to celebrate. This is a good opportunity to revise your case study and present your results to executives and engineering leaders once more. You’ll also work on the final set of requirements for a GA launch. Again, set a date and assign your PM the owner for ensuring success.

Training and Support Resources

  1. Application team survey. Here’s a link to a survey we recommend sending to application team owners to determine if they are a good fit for the first install, Beta, or GA launch.
  2. Armory Spinnaker Documentation. We’ve developed administrative documentation (for platform teams) and user documentation (for application teams), which you can find here.
  3. Training. Armory provides two types of training for all of our customers: Administrative and User training. Administrative training should be attended by your platform team and includes information on how to operate, upgrade, and troubleshoot Spinnaker in production. User training should be attended by your application teams and provides guidance on best practices and how to use Spinnaker effectively. We also have a number of Spinnaker training videos here.

Learn More

Share this post:

Recently Published Posts

How Apple Uses Plugins for Continuous Deployment Success

Dec 7, 2022

Reaction to Apple’s Spinnaker Summit 2022 Talk At the most recent Spinnaker Summit, Joe Cavanagh and Benjamin Powell from Apple discussed how they maximize code reuse, eliminate repository maintenance, and unify their CI/CD process across many plugins. They also discussed the mutual benefits of a well-maintained organizational plugin ecosystem for Spinnaker users, developers, and operators.  […]

Read more

Release Round-Up – Armory  Continuous Deployment Self-Hosted / Managed 2.28.1

Nov 28, 2022

Welcome to the latest release round-up for Armory’s Continuous Deployment Self-Hosted and Managed (CDSH) solution. In this release, we’ve focused on a few quality of life improvements and bug fixes, as well as introduced two user-requested early access features behind a feature flag.  Here are the important quality of life and bug fixes in the […]

Read more

Better, Faster, More Reliable Deployments

Nov 22, 2022

Armory brings cross-environment orchestration, validation, and advanced deployment strategies to AWS Marketplace How fast is fast enough when it comes to deploying your software?  Deployment frequency, along with lead time to changes and rollback failure rate, is one of the top metrics of DevOps success. Armory is working to solve the DevOps deployment slow-down through […]

Read more