University of Sydney Cloud environment provisioning and containerisation using AWS CDK, CodePipeline and CodeBuild

The University of Sydney is a public research university located in Sydney. Founded in 1850, it is the oldest university in Australia and is one of only six sandstone universities nationally. It was one of the first universities in the world to admit students solely on academic merit, and open their doors to women on the same basis as men.

The university comprises eight academic faculties including Arts and Social Sciences, Engineering, Medicine and Health, Science, Architecture, Design and Planning; Sydney Conservatorium of Music, Sydney Law School, and University of Sydney Business School.

The Engagement

The student information system used by the university was an off-the-shelf solution managed by a vendor. The university was in need of additional non-production environments to allow the internal technology team to develop, test and release changes more frequently.

The vendor’s managed environments were expensive, difficult to provision, and required an extensive lead time. In line with the DORA metrics (deployment frequency), the university decided to speed up the development and release cycle by building the capability to provision additional non-production environments in-house.

The software in place was designed to run on traditional servers, but the university wanted to move away from legacy, and adopt more modern technology approaches like containerisation and managed services. This would achieve the dual goals of reducing infrastructure cost, as well as management overhead. Automation was also a key objective of this work. The university technology team is relatively lean, so it was necessary to automate environment provisioning to maximise speed and minimise effort and overhead.

The Solution

Midnyte City began the engagement by scheduling and facilitating workshops with key stakeholders in order to better understand the requirements, drivers, and constraints to be considered during solution design. Next, the Midnyte crew worked with the university’s technology team to devise a proof-of-concept to build the environments using ECS fargate, which would be key to securing vendor support for the initiative.

Upon successfully demonstrating that the application could run on containers, the next step was formulating the delivery roadmap in collaboration with key stakeholders from the university’s technology team, as well as the vendor. The team conducted regular showcases with key stakeholders to keep them informed of progress and elicit feedback and guidance.

Ultimately, the solution was building the containerised application using AWS CDK, CodePipeline and CodeBuild. These tools aligned with the current technology stack and the existing skills and experience in the university’s technology team. This way, we ensured that the solution sustains long after Midnyte City has completed the engagement. The final step before concluding the engagement was to plan and conduct knowledge transfer sessions, with accompanying documentation and recordings so the technology team can refer to these resources as needed.

The Results

Fast environment provisioning
With the implementation of automated pipeline and infrastructure as code, the university’s technology team is able to respond to new environment build requests quickly and easily. With a simple configuration file change in the infrastructure code repository, the pipeline handles all the provisioning and de-provisioning of environments. This has reduced the environment provisioning time from weeks to hours.

Reduced running cost
As the environments are only required during business hours, a mechanism was established to automatically spin up and shut down instances (the operating hours are customisable through tagging). This reduced the running cost of the environment by 70% in comparison to running them 247.

Reduced maintenance overhead
‍Because we chose to run ECS fargate, there is no longer an ongoing requirement to maintain the operating systems and infrastructure running the container. This reduces the time and effort the technology team spends on maintaining the infrastructure, as well as the risks involved in having to manage security vulnerabilities in the underlying OS.