Projects – HarishCoorg

HPE OneSphere

HPE OneSphere: Onboarding Kubernetes on Bare-Metal

HPE OneSphere was a cloud management platform offered by Hewlett Packard Enterprise (HPE). It was designed to simplify the management of hybrid cloud environments, which typically involve a combination of on-premises infrastructure and resources hosted in public and private clouds.

Project Goal`

Simplify and streamline the onboarding of Kubernetes clusters on bare-metal within HPE OneSphere, reducing manual intervention and improving admin confidence.

The problem

Deploying Kubernetes on bare metal was:

Complex, with too many manual steps.
Unclear (admins struggled to identify master nodes).
Time-consuming, delaying operations.

This caused delays, errors, and reduced adoption of Kubernetes in OneSphere.

My Role

UX Research & Persona Refinement
Competitive Analysis
Task Flow Mapping
Wireframing & Prototyping
User Testing (Preference, Moderated, A/B)
Final UI Design

Research & Insights

Conducted stakeholder and user interviews.
Benchmarked competitor solutions (VMware Tanzu, RedHat OpenShift, GKE, EKS, AKS).
Key insight: Competitors offered automation and transparency, while OneSphere’s process felt manual and unclear.

Persona

Task flow

Previous flow: Multiple manual steps → error-prone.
New flow: Automated node attachment + clear master node designation.

Competitive Analysis

Company Name	Product Offering	Unique Value Proposition	Features
Hewlett Packard Enterprise (HPE)	HPE OneSphere (Cloud Management)	Simplified hybrid cloud management and Kubernetes orchestration.	Multi-cloud management Unified interface Cost management Resource provisioning Analytics and insights Security and compliance
VMware	VMware Tanzu (Kubernetes Platform)	Comprehensive Kubernetes management and development platform.	Kubernetes cluster provisioning Application deployment and scaling Monitoring and logging CI/CD integration Multi-cloud support
Red Hat (now part of IBM)	OpenShift (Kubernetes Platform)	OpenShift (Kubernetes Platform)	Kubernetes orchestration Developer tools and APIs Security and compliance CI/CD pipeline integration Extensive partner ecosystem
Google Cloud	Google Kubernetes Engine (GKE)	Managed Kubernetes service on Google Cloud Platform.	Automated cluster management Seamless integration with GCP services Scalability and high availability Kubernetes Engine Marketplace
Microsoft	Azure Kubernetes Service (AKS)	Managed Kubernetes container orchestration service on Azure.	Integrated with Azure DevOps Monitoring and diagnostics Enterprise-grade security Serverless Kubernetes with Azure Functions
Amazon Web Services (AWS)	Amazon Elastic Kubernetes Service (EKS)	Managed Kubernetes service on AWS.	Automatic updates and patching Integration with AWS services Scalable and highly available AWS Marketplace for Kubernetes applications

Wireframes & Testing`

User Testing

Conducted multiple rounds of user testing

First, we did preference testing to understand which design is better than the other designs
Made some modifications based on the feedback received from stakeholders
Conducted moderated user testing on the new design to understand what can be made better and whether users are able to relate to the work they perform
Finally we developed both the options and did a AB testing.

Following are some of the findings from these tests

Users tested

Users felt its the easiest method

Users unable to find which nodes were master nodes

mintes to complete the task

Few Verbatiums from the test

Attaching nodes was relatively quick and easy. I had no issues with figuring out how to go about doing this.

I believe that this would be the absolute easiest way to attach nodes, It is quicker and much more efficient than the previous method.

I believe all three nodes were attached as a master node to the cluster. I could be wrong but this is what I can recall

Even though the process was easy for the users, 2 users were unable to understand which node was attached as the master node. So we went back to the drawing board and made some changes to address that and tested it again.

Final Designs

One-click node attachment.
Automated master node creation.
Clear indicators for node roles.
Simplified update flow.

Impact

Time savings: Reduced onboarding from ~6 min to ~2 min.
Confidence: 92% of tested users felt “easiest method yet.”
Adoption: Increased frequency of successful Kubernetes deployments.

Reflection

This project pushed me to:

Translate deeply technical workflows into simple, intuitive designs.
Balance automation with transparency, ensuring admins felt in control.
Leverage testing insights to iterate quickly and land on a solution that worked in the real world.

If I continued, I’d explore:

AI-driven recommendations for node assignment.
Deeper monitoring integration to further reduce admin effort.