Private cloud.png

Deployments - Private cloud

Deployments - private cloud

Team - Agoda engineering
Platform - Web
Role - UX designer with 1 technical product owner, 4-5 software engineers, 1 front-end engineer
Duration - 6 months - Ongoing


What are we trying to solve?

The engineering team is aiming to transition away from the outdated legacy deployment platform known as Cspider.

Some issues faced are

  • It's hard to add new features to a large, complex codebase.

  • Doberman (In house alerts) and canary deployments are not working properly.

  • Onboarding and resource requesting are disorganized and challenging.

  • It is hard to see what was deployed, where, and any errors.

    Their goal is to adopt private cloud deployments using Kubernetes, with a strong focus on enhancing the overall developer experience throughout the deployment, monitoring, and management processes.

    It also helps them with resource management overtime and reduce operational cost of virtual machines.


What did we do?

It was evident when i joined that the team was behind timeline and wanted to migrate all services to private cloud soon. These are the actions i took,

Our process here includes

  • Understanding the domain/User interviews

  • Prioritisation and alignment

  • Pragmatic solutioning

  • Feedback and iteration

  • Presenting designs to tech leadership


Understanding the domain

As a product designer, I seek to understand developers' work processes and sought insights from experts to better understand challenges they face.

Snippet of using dovetail to consolidate findings

Prioritization and alignment with engineering team

Main findings - Needs

  • Access to documentation easily as all engineers are expert Kubenetes users.

  • Have the ability to continuously or manually deploy releases.​

  • See the history of my application and deployments.​

  • Rollback my service in the event of any problem.

  • Manage resource allocations for my service  (and self service)​

  • Be able to see resource usage, metrics and the current health or status of my service.​

  • Distribute deployments across multiple clusters and DCs. ​


Solutioning

Before - Legacy platform

After - Deployments on Developer portal


Success and outcome

  • First set of service migration was successful. We measured the experience of migration through a set of surveys.
    4.74/5 Sample size 35 teams

  • Over a full year design pushed for and improved the system usability scale (SUS) score for the platform from
    22Q4 - 61.9
    23Q2 - 63.375
    23Q4 - 68.825


Reflections

As the sole designer working in this project. It was very fulfilling and challenging working on a new domain area.

  • Proving design value past “getting the job done” in an engineering team is extremely challenging.

  • Find allies in the engineering team that can help you push ideas to the organization.