Leveling up data analytics for the cloud era
Customers who wanted to run Teradata analytics on their own AWS or Azure databases had to file a ticket in Teradata's ServiceNow portal and wait for staff to work it by hand. Turn-times were slow, and as the market moved toward self-service tools, that ticket queue started costing Teradata customers who didn't feel the company was keeping up. I led visual and interaction design for Intellicloud, the console built to replace it, working with two product managers and a team of engineers over six months.
A ticket queue instead of a console
Provisioning or managing a database on AWS or Azure meant filing a ServiceNow ticket and waiting for Teradata staff to work it manually. Turn-times were slow, and the market had already moved toward self-service, easy-to-use products. That gap was starting to cost Teradata customers who didn't see the company meeting their needs.
I came onto the project after initial discovery, where the team had confirmed real demand for integrating public cloud databases, just no product to do it through. The goal became a single, self-service console: provision and manage a database in minutes, not a ticket.
The ask wasn't to invent a new product category, it was to bridge two separate services, AWS and Azure, into one place customers could provision and manage databases themselves. I toured how AWS, Azure, and Kubernetes each handled the same job before deciding what our version needed to focus on, then built the interface on Covalent, Teradata's early, loosely defined Material Design pattern library, extending it into a coherent system as the most significant project built on it to date.
Dashboard
The primary point of entry: every database site, its provider, storage, and utilization, all from one screen.
DBAs could switch between sites, provision a new one, and drill into status, backups, and security without leaving the page.
Provisioning a site
Provisioning was a multi-step decision, not a single form. Platform, region, database version, and product tier each changed what came next, so I erred on the side of a guided wizard. Patterns like this are sticky, and used correctly they lead to better completion than dropping every option on one screen.
Backups
Customers needed to trust their data was protected without babysitting it. The backup wizard let someone select databases, choose a segment, and schedule recurring jobs, while the dashboard's Backups panel showed job history and status without a support ticket.
Data shipping
This wasn't part of the original scope. Near the end of the project, Teradata asked if I could design a vision for setting up, tracking, and managing the transfer of data from on-premise systems into a cloud instance. The result carried the same multi-step pattern as provisioning and backups, so it read as part of one console instead of a bolted-on feature.
Metering
Customers needed to see what they were actually using, by database and by month, to manage cost and capacity themselves instead of asking Teradata for a report.
Getting to one system
I came into database management as a novice, so I asked our technical SME to walk me through what a DBA might already be using. AWS tried to show everything in one interface. Azure kept elements in view with a window system for context. Kubernetes consoles pointed toward a more considered Material UI for desktop. All three had immense functionality; the question was what our version actually needed.
Scope came from a story-mapping workshop with Teradata's two product managers, who hadn't documented the functionality they wanted before I ran it. Story-mapping isolated the core jobs, broke them into tasks, and showed the dashboard had by far the most functionality, so that's where design exploration started.
Teradata wanted to see iterations in high fidelity from the start, so alongside scheduled reviews I kept comps flowing to my day-to-day stakeholders asynchronously. Earlier passes worked through unclear primary actions and inconsistent status colors before the shipped version settled into one coherent pattern.
From concept to shipped
The dashboard evolved through iteration and discussion with Teradata's PMs. Early passes tested how far a card-based layout and status color could carry the primary action before the design settled into the table pattern that shipped.
Additional explorations
With feature work delivered, I looked at problems that hadn't made it into v1: a systems-ops view of tickets and customer health, and applying the evolved pattern library to Zeus, a separate internal query tool a colleague had already built for Teradata.
Future vision
The ops dashboard closes the loop back to the ServiceNow queue this project replaced for customers, giving Teradata's own team the same visibility on the other side. Zeus wasn't mine to design, but restyling it showed how far the pattern library could travel beyond the console it was built for.
Impact
The Intellicloud Management Console launched in late 2017 with AWS support, and Azure followed in early 2018. My design and engineering work was handed to a new internal Teradata team for continued growth, and the console became part of what's now Teradata Vantage.
- Shipped fast enough to show customers Teradata had heard them and was responding, slowing the losses the ticket-based process was causing.
- Built a modern, services-focused infrastructure that reduced Teradata's operational costs, cut time to market, and enabled continuous integration for faster delivery going forward.
- Teradata had no team to conceive, build, and manage front-end software before this project. We helped vet and hire the people who took it over, and left them processes and patterns to keep evolving it.
Accomplishments
- Led visual and interaction design for the Intellicloud Management Console end to end, on a team of one designer, two product managers, ten engineers, and two QA engineers.
- Researched AWS, Azure, and Kubernetes consoles, then ran a story-mapping workshop with product to scope what the console needed to cover.
- Extended Covalent, Teradata's early Material Design pattern library, into a coherent system across provisioning, backups, data shipping, and metering.
- Designed a vision for data shipping and an ops-facing dashboard beyond the original scope, at Teradata's request.
Reflection
Developer tools don't have to be messy. Given a bit of structure, they make navigation and task completion easier, even for something as dense as database provisioning.
Organizations feel shame about change the same way people do, and transformation is hard. Getting Teradata comfortable meant early iteration and frequent presentation, building their confidence in a direction before asking them to commit to it. If I did it again, I'd push harder for direct customer inclusion in the process itself, and I'd have codified the patterns I built back into the Covalent framework team rather than leaving that for whoever came after me.