Bryte helps Abacus International improve business intelligence and analytics capabilities to understand travel patterns and customer buying behaviour.
Abacus International is the leading provider of Global Distribution Systems to the travel industry in Asia Pacific, serving over 100,000 travel agents across 59 markets. The business processes more than 1.1 trillion transactions each year, providing a rich source of data that can be explored to uncover business insights.
Abacus wanted to improve business intelligence and analytics capabilities to understand travel patterns and customer buying behaviour in near real time using a multitude of data sources like online ticket sales and passenger itinerary data.
- Scale: The current on-premise infrastructure was difficult to scale up and struggled to keep up and provide high performance with growing data volumes.
- Data Silos: Large amounts of data existed in multiple sources and structures.
- Speed: It took up to 8+ hours of overnight batch workloads for existing on-premise databases to process the information and deliver required business reports.
Bryte worked with Abacus to deploy Amazon Redshift for one of their key markets in Asia Pacific (Korea) as a test case to look at performance and scalability benefits.
An automated process was set up to continuously sync Korean market data from the existing on-premise Oracle Databases with Amazon Redshift. This enabled new data to be available in Amazon Redshift near real time. Our team then developed a range of highly efficient ELT processing jobs to wrangle multiple data sources and build rich Data Marts and Business reports. Then, our team conducted AWS performance monitoring and optimisation to deliver the following results.
- All key reports could be refreshed, updated and reproduced in under 10 minutes providing Abacus Executives with the capability to react to market conditions as they occurred in near real time, delivering unprecedented insights and agility.
- Immediately after migrating existing workloads to AWS, there was at least a 2x performance improvement on the same reports. For example, a report that used to take 45 minutes to run, could now be executed in under 20 minutes.
- Through further ELT processing, data modelling and Amazon Redshift performance optimisation, processing times were further reduced by a factor of at-least 5-10 x. Reports which ran in 20 minutes were executed in under 2 minutes. Other reports that used to take hours to run, were now generated in under 10 minutes.