Assumption 1
Monitoring cloud health is the most important part of a cloud manager’s daily job.
A heat map would show IT managers the cloud health status effectively and accurately
Project Overview
AirDeck Technologies builds the next generation private cloud solutions to small and medium business. As the primary user, an IT manager needs to monitor the health of it’s cloud system and triage accordingly when there’s an issue on the daily basis. My stakeholders were very keen in building a dashboard to help IT managers make informed decisions. As the UX designer for this project, I focused on researching, learning, validating assumptions.
My Responsibilities
- Research & Discovery
- Workshop & Ideate
- Test & Iterate
Tools
- Whiteboard
- Sketch App
- InVision App
- UserInterviews.com
Problem Statement
IT manager needs a area inside of the cloud control panel to monitor cloud performance, energy consumption, receive alerts and triage issues, etc. With a wide range of performance data available at any given time, how can we present the most effective and efficient data to support their daily decision making?
Goals
Research on the critical information the cloud managers are looking for and what type of critical tasks they perform. Test preliminary designs, gain insights on our assumptions and iterate the design solutions to meet the cloud manager needs.
Design Process
Research
- User Interviews
- Task Analysis
- SME Conversation
Ideate
- Design Workshop
- Storyboarding
- Role-Playing
Design
- Wireframe
- Design Mockup
Test
- Feedbacks
- Prototype Testing
- Design Iteration
User Interviews, Task Analysis & User Story Mapping
I interviewed 8 cloud managers and facilitated task analysis and user story mapping activities. The goal of the research is to paint a picture of what a cloud manager’s typical work day looks alike. What type of work do they spend the most time in? What are the most important tasks on daily basis? What are the essential information they need to perform the most important tasks?
Subject Matter Expert Conversation
To better understand the users of a cloud management application, user story mapping, storyboard and idea pitch workshops were introduced. The purpose was to build a shared understanding on dashboard design assumptions made by the SMEs.
Assumption 2
Cloud managers spend most of their time in tracking system alerts.
An event notification chart would allow IT managers to respond to a cloud anomaly quickly and act accordingly
Assumption 3
Capacity changes over time are important metrics.
A cloud capacity chart is needed for IT managers because it can forecast cloud performance and manage the future capacity accordingly
Assumption 4
An energy consumption chart is needed for IT manager to better manage resources and environmental impact
Sketches & Ideations
Based on the assumptions above, I produced design concepts for the dashboard. The purpose is to show these ideas and receive feedbacks from IT managers.
Validate & Invalidate Assumptions
Dashboard designs were presented to 8 IT managers. I was particularly interested in finding the answers in the following questions:
How participants respond to the cloud condition heat map in the way it’s presented.
How participants respond to the historical data such as capacity and ennergy consumption.
Does the dashboard present the most critical information they’re looking for.
Findings & Insights
There’re a lot of interesting discoveries from the feedback sessions. Some of them really challenged my thoughts around the “Context” and “Meaning” of the dashboard charts. The cloud heath map was overwhelming and confusing to most of the participants . Participants preferred a simpler, more accurate and straight forward way to display cloud health.
Historical Data
4 out of 8 participants find showing historical data on dashboard is confusing. In their views, historical data such as energy consumption and capacity is not the most pressing information they look on a dashboard on daily basis.
Current Data
The current data such as cloud heath at Node, Virtual Machine, and Virtual Network is vital to their daily responsibilities. An effective alert system is another essential tool for them to monitor and react to the cloud conditions.
View Only Alerts
View only alerts show when, where and what have happened in their system, but they don’t show how and why.
Clustered & Actionable Alerts
Clustered alerts can better show severity and correlation of the events. The ability to take immediate actions after receiving a critical alert is also crucial to the user.
Design Iteration
Based on the insights gained from test participant feedback. I iterated my dashboard design with some major changes. The cloud health heat map created visual noise and confusion for the participants. It was removed and replaced with accurate component data presentation. The Critical Alerts chart focused on showing alert correlation and severity. It works side by side with alert table to proved the user actionable insights: which alert should I respond to and what are the available triage actions to take.
Dashboard now only shows current data, no more historical data.
Accurate and easy to read component data
Alerts table can be filtered by the event cluster bubbles.
Recommended Actions are provided for trouble shoot.
Critical Alerts are presented in cluster bubbles. The size of the
bubble indicates the number of the alerts grouped in each cluster. A
cluster bubble show there’s a strong correlation among those alerts
which could be traced to a common cause.
Impact, Results & Next Steps
My revised dashboard design received positive feedback from the test participants after another round of testing. The accurate component data is easily to read and comprehend. The combination of the alert status graph and alert table provided two different ways to monitor alerts: 1. Using graph to show context, correlation and severity. 2. Using alert table to organize, pinpoint and triage anomalies. The test participants found they provide comprehensive, multi-dimensional insights during alert responding and triaging tasks. My stakeholders decided to build an MVP based on the revised dashboard design. I intended to research and evaluate the new design in another round of test.
Back to UX Work