T-Mobile Supports 5G Rollout with Azure Synapse Analytics and Power BI

Building a nationwide 5G network

T-Mobile is one of the largest providers of wireless communications services in the United States with over 100 million customers generating tens of billions of dollars in revenue annually. The organization has experienced huge growth over its nearly 25-year history, and in 2020 completed a merger with Sprint, another American telecommunications company. Headquartered in Bellevue, Washington, T-Mobile is focused on providing customers with its unmatched combination of value and quality, unwavering obsession with offering them the best possible service experience, and undisputable drive for disruption that creates competition and innovation in wireless and beyond.

Over the past few years, T-Mobile has focused on delivering 4G LTE while building a nationwide 5G network. Executing a fully functional 5G network nationwide required a massive ramp up in cell tower site build with the enterprise seeing an over tenfold increase in the start of weekly construction projects. To ensure that these sites were being built on time T-Mobile optimized their supply chain and ensured efficient business efforts. Effective and accurate reporting on data and metrics related to supply chain and business critical information became increasingly crucial to the success of T-Mobile’s 5G goals.

As T-Mobile began putting together reports to help cover build progress and supply chain metrics, they quickly realized that the current landscape of disparate data sources and systems made it difficult and time consuming to ensure real-time metrics. Pulling data from all the necessary sources took over 12 hours. The team needed a solution to centralize data to optimize planning and reporting that would be effective and efficient. Members of technical staff, Robert Thompson and Geoff Freeman, led this initiative as part of the Data Solutions and Analytics division at T-Mobile.

Creating a lakehouse solution with Azure Synapse

The first step for T-Mobile was to centralize all their data. The team's initial solution was using Microsoft Azure Data Warehouse, internally known as TMUS. While TMUS created a centralized source of data and made improvements on the speed to insights, the system still didn’t meet all the needs of T-Mobile site engineers. Data was not processed quickly enough, and report access wasn’t well-defined, leading to more reports being modified than necessary. Migrating new data sources into TMUS was not seamless and prohibited business units from getting quick access to all the data they needed to write their own transformations.

Looking to improve the features and performance of TMUS, T-Mobile decided to look at a data lakehouse—a hybrid approach that combines features of a data lake and a data warehouse. With a data lakehouse security would be improved while eliminating workload contention and ensuring data isolation. Data would easily be added and centralized through the serverless architecture. By choosing to build their solution with Microsoft Azure Data Factory, Azure Synapse Analytics, and Azure Databricks, data within the lakehouse would be significantly more accessible and flexible across the organization. Plus, migrating reports and dashboards would be easy with the help of Azure Synapse Analytics. Serverless SQL endpoints within Azure Synapse Analytics easily enabled data and reports to migrate effortlessly. Because compute for Serverless SQL endpoints is significantly less expensive, T-Mobile was also able to save money on the Dedicated SQL pool when users queried the lakehouse.

After deciding on using a data lakehouse, T-Mobile began the process of revamping their existing architecture. The ingestion level in the new architecture remained primarily the same since most of the extraction processes were staged in the data lake, but to help ingest and shape the data a series of delta tables needed to be added.

Once the delta tables were incorporated, the team solidified their foundational architecture. Data gets ingested through Azure Data Factory from Oracle, MySQL, SQL Databases, SAP HANA, and other locations. Data is then ingested into the Delta Lakehouse through a series of delta tables: Bronze (raw data), Silver (refined data), and Gold (curated data). Using Microsoft Power BI, users then connect to the Gold data using a Serverless SQL endpoint. Data is also pushed into a dedicated SQL pool for legacy processes that need it. With a Serverless SQL endpoint within Azure Synapse Analytics, data can be exposed in such a way that makes it easy to migrate T-Mobile’s existing reporting ecosystem and enable the data lakehouse vision to come to life.

Bringing supply chain data to life with Power BI

When it comes to reporting, Power BI is the main tool used within T-Mobile. Teams across the organization are using Power BI Premium to power their analytics and data-driven decision making.

For T-Mobile, Power BI’s seamless integration with the Azure ecosystem and reduction of costs made it the obvious choice as their analytics tool. Plus, with secure file access as a high priority, Power BI’s integration with Azure Active Directory and authentication pathways ensure that users could be added to an active directory group and T-Mobile can continue to maintain their high security standards. Azure Synapse Analytics also helps manage and regulate permissions-based roles and credentials. Alongside Azure Active Directory and Azure Synapse Analytics, T-Mobile also uses Azure Security groups.  “With Azure security groups, [managing security] became a much easier task. Managing permissions … is way easier to manage than going through a traditional security portal” says Vardhaman Patil, Senior Business Intelligence manager. Leadership at T-Mobile remains focused on creating a data driven environment and leveraging Power BI to champion that initiative.

When it comes to procurement and supply chain reports and dashboards, supply chains are broken down into two avenues: retail supply chain and network build supply chain. The retail supply chain manages store inventory and the supply and demand surrounding storefront and online retail needs. The network build supply chain covers all the materials and assets needed for network site builds. Patil, alongside Bruce Trinh, Business Intelligence Manager, focused on building reports and dashboards for the network build supply chain to support the rapid cell tower site production.

Over the past two years, Patil and Trinh built 30-40 reports supporting the network site builds, reports spanning planning to reverse logistics. Reports also have a wide diversity of users, with reports being used by leadership to make strategic decisions to warehouse managers leveraging operational reports.

The fulfillment report is a core report to the network build supply chain. The report offers high level overviews of the status of warehouse inventory, the health of supply chain, tracked packages, and more. The report has multiple tabs to enable users to toggle through different filters and features to find the specific data they are looking for. The report uses DAX to calculate some of the metrics, but most of the report logic is sorted out in the backend to keep Power BI reports as efficient as possible.

Another high traffic dashboard is the Kit Management dashboard. Supported by the embedded Power Apps functionality, the Kit Management dashboard helps build sites track the kits being used, and the transit of kits to and from sites.

With the data lakehouse architecture, accessing reports and real-time data is significantly quicker. Once data was made available in the delta lake, it became significantly easier to switch Power BI reports via Serverless SQL endpoints. Users can schedule updates and manage refresh cadences to ensure that the data in reports fits their needs. Reports are accessed through a centralized catalog that houses links to all the reports supporting network site builds. The reports are shared across the entire organization via premium capacities.

Embracing the data lakehouse and encouraging self-service reporting

Over the past two years of building reports, Patil and Trinh’s team, with the help of a vendor team, built team-wide reports that are used by business partners and individuals working on Network Builds. Now, while the data team occasionally builds reports to help specific verticals, a new internal team has been dedicated to handling the longevity of T-Mobile’s reporting platform and setting up new premium workspaces.

After seeing the cost savings and reliable performance of their new data lakehouse solution, T-Mobile is looking to continue developing and expanding the architecture as well as self-service reporting capabilities. Working to increase the automation of data quality and the discoverability of data, the data team hopes to support business teams as they begin to learn and feel empowered to build their own reports and use data to make crucial business decisions.

For more information: LinkedIn, Twitter, Facebook