Case Study
UST IQ transforms supply chain analytics for a global retailer with a 50% reduction in operational TCO
CLIENT
This global retailer operates a network of approximately 10,000 pharmacies, health, and wellness stores across multiple countries. As a leading provider of prescription drugs and health-related products, the company offers services that include retail, wholesale, and distribution of pharmaceuticals. Through both in-store and digital channels, the company leverages technology to offer innovative services, like digital health platforms and prescription management tools.
CHALLENGE
Existing data infrastructure couldn’t handle massive data volume, hindered insightful analytics
The client faced significant challenges in managing and analyzing its vast business data repository efficiently. The company’s sales history table held over 700 billion records, while its item master included over 100,000 unique items spread across a network of more than 9,000 stores. This immense data volume made it increasingly difficult to perform analytics needed for effective business decision-making.
The company’s existing Hadoop and MapReduce implementation ran on a 500-node cluster, often requiring more than two weeks to complete certain complex queries and analysis. This prolonged processing time prevented company leaders from gaining timely insights into sales patterns, customer behaviors, and inventory needs, which are crucial for informed decision-making and responsive business strategies.
The company attempted to implement Apache Spark as an alternative. However, the Spark setup introduced significant memory management issues due to the scale and complexity of data, resulting in frequent system crashes and unpredictable performance. Additionally, the operational cost of running Spark at this scale was unsustainable; thus, the solution failed to solve the problem. These inefficiencies strained the client's IT resources and increased operational expenditures, diverting budget dollars from other critical areas.
The client required a robust solution that could handle the substantial data volume and provide interactive query capabilities, allowing analysts to obtain near real-time insights. The solution also needed to be scalable and adaptable, capable of expanding as data volumes continued to grow.
Without a more efficient, scalable, and cost-effective approach, the client risked falling behind competitors who were better equipped to leverage big data for strategic advantage. This ongoing struggle with data latency and processing costs highlighted the need for a transformative solution that could support the company’s need for large-scale analytics with improved performance and financial viability.
TRANSFORMATION
UST IQ deployed on existing infrastructure transformed supply chain analytics
UST delivered a comprehensive supply chain data analytics solution by deploying UST IQ on the client’s existing Hadoop cluster, leveraging the existing architecture to efficiently handle the immense scale of data. UST IQ introduced a shift in the design paradigm from relying solely on in-memory, on-demand data processing to a pre-compute and refresh model.
The new approach allowed the system to pre-compute key analytics and metrics at scheduled intervals, significantly reducing the need for real-time, on-demand compute power for each user query. By calculating and refreshing insights periodically, UST IQ ensured that crucial data was readily available when needed without overwhelming compute resources, which optimized operational costs.
To further enhance efficiency, UST reserved in-memory processing for batch operations conducted during off-business hours. This setup allowed the client to leverage powerful in-memory analytics without compromising system availability or increasing memory overhead during peak operational hours.
The UST IQ architecture also minimized latency by ensuring that pre-computed data could be retrieved instantly for common queries, providing near real-time insights for time-sensitive business decisions. This strategic allocation of resources not only maximized compute capacity but also prevented costly memory management issues previously encountered with Spark.
The pre-compute strategy eliminated the need for extensive parallel processing and on-demand scaling, substantially reducing costs. UST IQ also provided built-in scalability, allowing the system to handle growing data volumes without requiring additional hardware clusters or financial investments.
With UST’s solution, the client can now perform complex analytics efficiently across its 700 billion-record sales history, achieving results in hours rather than weeks. The solution also allowed for timely data refresh cycles, ensuring that analytics remained relevant and up-to-date as new data streamed into the system. Analysts can take advantage of interactive query capabilities over its large data set to enhance strategic decision-making.
IMPACT
From weeks to sub-second query processing at half the total cost of ownership
By implementing UST’s solution, the client achieved significant improvements in its data analytics capabilities, particularly around query response times and overall cost efficiency. The most notable improvement was sub-second queries—a dramatic improvement over the previous system that took weeks to deliver complex analytics results. This instant access to insights allowed business teams to make data-driven decisions faster, enhancing business agility to respond to market demands and operational needs.
The dramatic improvement in query times enabled the client to streamline supply chain operations with real-time data insights with more accurate forecasting, optimized inventory management, and streamlined logistics planning. With quicker access to sales data and inventory levels across the company’s 10,000+ stores, leaders could identify trends and shifts in demand almost immediately, optimizing stock levels and minimizing lost sales due to stockouts. The new analytics capabilities also supported more responsive pricing strategies, helping the company adjust prices dynamically based on recent sales trends, customer demand, and regional performance.
The solution also resulted in a 50% reduction in the total cost of ownership for the company’s supply chain analytics operations. By leveraging pre-compute models and off-peak in-memory processing, UST IQ drastically reduced the need for continuous high-performance computing power, cutting operational costs in half. This approach eliminated the need for constant on-demand scaling and high-memory nodes, minimizing its infrastructure footprint without sacrificing performance. The shift to a more cost-effective model also freed up budget for further innovation, allowing the client to invest in new capabilities and improve other business areas.
The enhanced analytics solution also reduced dependencies on external consulting for data analysis since in-house teams can now access and interpret data-driven insights independently. This autonomy not only cut down costs but also improved data security because sensitive data remained within the client’s ecosystem.
This data analytics transformation provided a scalable solution for the company to meet its data growth needs and evolving business demands, ensuring sustained competitiveness in the rapidly evolving retail landscape.
If your organization is struggling with data analytics, UST IQ can help you. Learn more here.
RESOURCES
https://www.ust.com/en/boundless/supply-chain
https://www.ust.com/en/what-we-do/digital-transformation/data-analytics