?

University of Phoenix Realizes Benefits of DevOps

By migrating applications around DevOps paradigms, the University of Phoenix has streamlined IT operations and is now deploying services faster than ever.

About the University of Phoenix

Founded in 1976, University of Phoenix helps adult learners move efficiently from education to careers in a rapidly changing world. Flexible schedules, relevant courses and interactive learning help students more effectively pursue career and personal aspirations while balancing their busy lives. The university serves a diverse student population, offering degree programs from campuses, resource and learning centers across the U.S. as well as online throughout the world.

Challenge

Modern day universities face many of the same IT challenges that large enterprise organizations struggle to overcome. One of the most common and difficult to address challenges is application management. Traditionally it’s been difficult for IT operations teams to accommodate multiple development teams — each with their own deployment requirements, toolsets and processes — while at the same time maintaining operational best practices, security and agility. 

The University of Phoenix faced similar problems around application management, specifically when it came to internally-developed applications from different software teams. The difficulty of maintaining operational controls, while allowing software development teams to move quickly, was becoming a scaling issue that needed to be addressed. 

Breaking that down into IT lingo 

  • Outdated deployment methods were causing long deploy times.
  • Patching monthly versus weekly left security and bug fixes unresolved.
  • Development cycles couldn’t be performed quickly, reducing agility.
  • More resources were being invested on infrastructure versus quality products.

Solution 

To lay the groundwork for solving these challenges, and potentially many others, the University made the decision to migrate to the Amazon Web Services (AWS) cloud. However, just moving to the cloud was not enough. IT operations realized that they had to change their operational paradigm, internal processes and be adaptable enough to work across different teams. None of those tasks are easy or accomplished in a short time frame. 

The plan

WWT cloud architects and consultants worked with University staff to collaboratively create a plan centered around DevOps paradigms and a common framework and toolset for operational consistency. This included:

  • A single platform on Kubernetes (K8s) to allow any developer to deploy any application they want without waiting for resources to be created for them.
  • CI/CD pipelines created with Bamboo and Harness to allow apps to run tests and deploy to the correct environment.
  • Rolling with additional batch and canary deployments to support requirements for each team.
  • Creating all the infrastructure in Terraform to take advantage of Infrastructure as Code (IaC).
  • Training sessions with development and operations teams around containers and K8s.
  • Training sessions with DevOps teams around Terraform to enable them to create and maintain infrastructure.
High-level diagram of K8s environment, orchestrated with Rancher and Terraform
High-level diagram of K8s environment, orchestrated with Rancher and Terraform.

 

Tools and services

Specifically, the solution used a combination of native AWS services and third-party tools, including: 

  • Auto Scaling Groups (ASG)
  • Elastic Kubernetes Service (EKS)
  • RDS Aurora
  • Elasticache
  • Elasticsearch
  • S3
  • Rancher
  • Terraform
University of Phoenix's CI/CD pipeline
University of Phoenix's CI/CD pipeline.

Results

Together, WWT and University of Phoenix teams deployed the above architecture into AWS and began the migration of production workloads into the environment. It was not long before the university saw the gains of implementing the new solution. Some of most impactful outcomes, included: 

  • 1-5 minute deploy times.
  • Seamless deployments with Bamboo and Harness that encourage developers to use CI/CD pipelines instead of doing it manually.
  • First ever 90 days in production with zero  application downtime.
Share this