In 2015, Inc. Magazine named Ultra Mobile the fastest growing private company in the United States. Incredible accomplishment for a team that literally started in a small office and had exploded over 1000x percent by their second year. The Telco industry was (and is) extremely volatile and growth in subscriptions and revenue isn’t possible without continually leading the curve in terms of the product you offer, the pricing you provide and the services that back your product. With a ton of energy focused on leading (not reacting) and cost optimization to drive the best value with the highest quality the amount on the chaos on the back-end systems was formidable. A large part of our success in being able to support the data-driven culture our executives demand was implementation of a completely revamped BI infrastructure. Our journey to developing this infrastructure should be instructive to any company seeking to gain significant agility in a relatively short time. In 2017, TDWI recognized the innovation in our BI architecture by naming Ultra Mobile the winner of TDWI Best Practices Award.
Critical Questions
Initially, our BI process was comprised of emailing Excel files back and forth, which worked perfectly well for a time. But as our user base started to grow, so did our data assets. As powerful as data can be, it can also be quite unruly and wild if it is stored across a variety of different technologies and systems, and that was the case with us. We were quickly amassing data in a Hadoop distribution, multiple cloud technologies, streaming data storage, and many other enterprise systems. We started to find it difficult to answer basic questions like “Where is our growth?” and “Why are customers leaving?”
Our initial plan was to leverage big data analytics to answer these questions. We implemented a data warehousing platform built on Hadoop, but this modern infrastructure was still unable keep pace with the business, because our IT team could not build application-specific capabilities that would enable us to analyze the data quickly enough, nor could they build fact or dimension tables before the business rendered them irrelevant.
Although revenues had increased, profitability was still elusive, as inefficient processes bled resources. We lacked a holistic view of our organization. Ultra Mobile’s leadership challenged our BI team to come up with a more practical, more sustainable BI architecture that would improve both the speed and the coverage of information asset delivery.
BI Modernization – The Virtualization Effect
The BI team established four use cases with which to test potential solutions:
1. The solution needed to establish a governance layer to enable data SMEs to present a cultured business data model to our organization. Through our siloed development style, we had created significant overlap in the underlying data attributes, and it wasn’t always clear which business rules had been applied.
2. The solution needed to accelerate our migration to a Cloudera data warehouse for data staging.
3. The solution needed to accelerate the performance of the Impala queries against the Hive structures in the data warehouse, as the performance wasn’t acceptable for driving the interactive dashboards the team hoped to deliver.
4. Finally, the solution needed to present data in a consistent, single view, as the blending and federation of data in the current BI tool was not mature.
The team settled on a combination of the Denodo Platform, Looker, and Tableau/PowerBI. The Denodo Platform provides data virtualization and governance capabilities, Looker is the primary delivery layer, and Tableau/Power BI provides the analytical tools.
Working in tandem, these technologies met all four of our core requirements, and provided additional benefits as well. The new infrastructure provides a stable, consistent dataset, with which all stakeholders can collaborate seamlessly, and this has led to the discovery of many opportunities that had been hidden in our data swamp.
Innovation for a Better Future – A Whole New World
One of the highlights of this project was the raw speed at which change could be effected. We only spent two weeks installing, configuring, and optimizing the Denodo Platform. In that short time, governance processes matured, and as new information opportunities surfaced, the governance team were able to quickly assess the enterprise impact of the new assets. Data can be quickly modeled in the Denodo Platform, security is smoothly assigned, and performance can be easily optimized through the available caching options before being presented to the BI tools (i.e. Looker, Tableau) for consumption.
Since security can be administered within the virtualization layer, databases are unencumbered by security processes, and the BI tools can be applied consistently without fear of a data compromise. Every one of our business users is now consuming data through the data virtualization layer. As we continue to make enhancements to the data architecture, the data virtualization layer abstracts business users from the complexity. By bringing innovation to data delivery through the Denodo Platform, our BI team was able to not only meet but exceed its original project goals.
Our BI team is now poised to transform from a cost center to a profit center, as they continue to find ways to monetize insight. In the future, we hope to leverage the Denodo Platform to deploy live metrics, such as customer lifetime value, directly into Web applications and dashboards, through RESTful data services. This would empower reps in our call centers and retail portals to always be able to prescribe the next best action for every customer.
Unleashing the Power of Data – If We Can, You Can Too
Data virtualization is a powerful technology for taming the wild data assets that can range across multiple enterprise systems. With data virtualization in place, business users can much more easily extract meaningful insights from the data, to fuel growth. Since it worked so well for us, I believe that every organization should give it an honest try.
- Managing Unprecedented Change with a Modernized BI Infrastructure - December 14, 2017