SQL Server on Hadoop inadequate

The immediate choke point in the project was Melco’s legacy SQL Server data warehouse on top of a Hadoop-based data lake. Although Hadoop was valued as a low-cost, highly scalable storage tier, SQL Server was too slow to sustain concurrent users, and the more data ingested, the slower it got.

• Performance became unacceptably slow when concurrent users tried to access the same dataset.
• Data loads were not only slow but often subject to manual input, and massive data growth affected performance and scalability.
• Because so little data could be analyzed at once, marketing campaigns could build on only a couple months of customer history, blocking full monetization of seasonality.
• The sophistication of analytics was inadequate for supporting the Customer 360 project’s goals.

Modernizing the platform with Yellowbrick

After reviewing several providers, Melco chose Yellowbrick Data Warehouse for its ability to transform the digital environment into a high-performance analytics powerhouse, while complementing Hadoop and existing tools and ensuring a smooth transition for end users.

Results Include:

• Loading complex dashboards and analytic data structures now take a few seconds instead of several minutes, with Yellowbrick providing headroom to add more concurrent users and easily scale to support user growth
• Faster access to new data, thanks to rapid ingestion and the ability to query new data immediately, so that analysts can work with the latest information
• 4X historical data (nearly 12 months) is available for analysis, allowing the company to build more targeted campaigns and monetize seasonality much more effectively
• Up to 300X faster queries, dramatically reducing the time to insight and accelerating the speed of decision making