CSU/Unisys dev-ops automation roadmap

Presenters

  • Milten Garia (MG), Manager of Integration and Web Services, CSU Office of the Chancellor
  • Anand Deshpande (AD), Cloud Architect, Unisys, CSU/Unisys Dev-Ops Automation Roadmap
  • Rajeev Singh (RS), Executive Transformation Architect, Unisys, CSU/Unisys Dev-Ops Automation Roadmap

Agenda

  • CMS History
  • Business Drivers, Challenges and Target Areas
  • Vision & Strategy
  • Automation Roadmap & Journey
  • Automation Tools & Proven Results
  • Business Case: Data as a Service

CMS History

MG: discussed pre-history of storage, pottery for storage of grain and other foodstuffs, and invention of the pottery wheel. “Machines are a way to outwit nature.” “We’re not reinventing the wheel, we’re inventing the pot.”

Started in 1998, Mission to provide efficient, effective and high quality service to the campuses, students, faculty and staff of the 23 campus CSU System and the Office of the Chancellor. All 23 campuses are on HR.

Business Drivers, Challenges and Target Areas

  • Align institutional and IT priorities, i.e. GI2025.
  • Lagging technology
  • Increased Services at lower cost
  • Campus autonomy, empowerment: we want campuses to have the flexibility to manage their own environment.
  • Data security and access: we’re constantly under attack, and we have Oracle patches not just quarterly, but all the time now.
  • Digital integration: being able to integrate new applications into your environment without interruption. We need to be able to streamline this process.
  • Data-driven culture: we’re at 500K students now, we have mountains of student and administrative data, so this is a priority for us.
  • Speed of service to campuses: upgrades are constant, we’re trying to manage this better.
  • Realize operational efficiencies

Target Areas for Automation

  • Agile delivery of instances: being able to provide better responsiveness.
  • Handle Peek Load during registration: cooperation for calendaring of resources.
  • Optimize HR
  • Campus self-service of daily requests: campuses can deliver faster in many cases.
  • Automating mundane & administrative tasks

Vision & Strategy

RS: This doesn’t need to be a complicated discussion, there is no one right answer to what we need. Our operating principle: keep things simple; we need to be focused on the outcomes. We want to build a self-service portal where we can ask for whatever tools we need, being cloud- and tool-agnostic. Options are important.

  • Cloud Ops
  • Approvers
  • Cost Management
  • DevOps Tools

Strategy for ROI: Productivity feeds > Speed of delivery feeds > Quality improvements

  • Priorities: nobody is “busy,” it’s just a matter of priorities
  • Targets: simple and well-defined
  • Communication: keep momentum and provide tangible results
  • Actions: win fast and fail even faster

Automation Roadmap & Journey

AD: I’ll discuss the following topics and hopefully keep you entertained 🙂

  • CMS Hybrid Cloud Deployment
  • CMS Peoplesoft as a Service
  • CMS Data as a Service
  • CMS Auto Burst Capacity in AWS
  • Fully automated CMS Private Cloud

Journey to dev-ops is easy early on, but as you build it up it gets more complicated. Ad-hoc/Scattered/Shell Scripts > Devising dev-ops strategy for automation & config management > PoCs w/multiple CM Tool i.e. ansible, puppet > Setting base dev-ops infrastructure ( application deployment tools > educating & training staff > transforming & writing code for automation > evolving as true dev-ops organization.

We chose Ansible as our dev-ops tool. We wanted to setup our infrastructure as code, automating using Ansible, Jenkins, GitLab, Right Scale, Dome9, ServiceNow, VMware, AWS, Delphix for DB virtualization, etc. This stack removes layers of human approval for many commonly-requested things.

Ansible benefits: simple maintenance with strategic setup. Power tools: automation toolbox, workflow, agentless approach. Simple: no scripting, low learning curve, human readable automation, create prototypes quickly. Applied across multiple IT Operation Areas: storage, infrastructure, applications and databases to create a larger scale of improved operational efficiencies and reduced infrastructure costs. Use cases include: cloud, infrastructure, database, and applications.

Automation Tools & Proven Results

  • PS PIA build 3-4 hours > 1 minute
  • Automated web server validation: 1 hours & 2 FTEs > less than 3 minutes/automated
  • LInux/WebLogic/Oracle Patching: Manual 12 hours > automatic 1 hour or less
  • Automated maintenance window 6 hours 15 FTEs> more work in 6 hours
  • More examples were shared, I can’t type that fast 😉

Business Case: Data as a Service

Shared an Ansible infrastructure workflow that includes many of the elements available.

RS: we’re providing these services to Brendan’s team, and he’ll talk more about this later in the conference with respect to system data, i.e. BI initially.

Q: you’re talking about making this a priority, how do you get a “lift” without impacting operations? RS: as we move along the project we want to 1) see how do the implementations while automating them and 2) find out where we really spend our time. We want to enable campuses to do most of the day-to-day stuff on their own. That’s an incentive for us to allow folk to work on the exciting things we currently don’t have the time to work on.

Q: is the Ansible tool built for managing AWS or Azure? AD: it allows you use different modules to manage those services. For most things like EC2, it has modules that allow you to do most common tasks.

Q: are you using in the cloud confirmation and terraform in conjunction with Ansible? AD: we’re using Ansible pretty much exclusively right now.

Q: what’s the itinerary looking like for pushing this out, i.e. training, tools, etc.? RS: we’re going to post “how we did it” articles, a service catalog, a set of guidelines for security, etc.

Q: do you use a 3rd party to audit/verify your service is running properly? AD: no, Ansible does this for us.