Realtime Business Visibility & Scalability through Cloud Migration

Executive Summary

The client, a Fortune 500 company, is a well-known pioneer in the hospitality and entertainment industry and has been known for its family cruise line business. The company used a revenue management system to manage the prices of cruise ships based on seasonality and demand. A dashboard was used to visualize advanced bookings done by customers.

The on-premise database used to house all of the revenue management system’s data. However, the infrastructure was not processing data reliably and was not scalable enough to meet the spikes and increased customer demand. This caused the company to lose business on the bookings and more importantly on their customer satisfaction scores.

Factspan addressed the scale, speed, and reliability by developing multiple ETL & ELT data pipelines to move their on premises data storage system to the cloud. The solution helped the client teams to capture demand properly, reduce workflows, measure performance effectively across all segments and countries, and accurately pinpoint cash flow issues.

About the Client

The client, a subsidiary of a large-scale global entertainment company, is a world-renowned leader in hospitality and entertainment and has been recognized as the top cruise line for families. The company focuses on delivering exceptional guest service and creating family-friendly fun onboard their cruise line and on their private island getaways. The company prides in inventing and updating its offerings to provide guests with a diverse selection of experiences.

Business Challenge

The demand for cruise vacations varies greatly throughout the year due to weather, holidays, school breaks, and economic issues. When customers reserve cruise tickets in advance, a revenue management system aids in managing price and booking – making reservations at the best possible price for a given timeperiod. It is also used to control the cruise line’s assortment of pricing options, including various stateroom categories, meal plans, and excursion packages.

Unfortunately, all the data from the revenue management system were stored in an on-prem database. At peak demands for advance bookings, the on-premises databases required manual intervention to scale, which became time-consuming and error-prone. The company started losing out on customers when the data infrastructure was not processing data accurately and was not scaling enough to meet the required demand.

Our Solution

Factspan built multiple ETL and ELT data pipelines to handle real-time data. Creating separate channels allowed business units to precisely get the data they need and in the format they require.

Prior to loading the raw data into a Snowflake data warehouse, the team first collected data from a central repository and moved it to an S3 bucket for staging. AWS batch jobs, at specified intervals or triggered by specific events, were used to perform data transformations on the source data in the Amazon S3 bucket, using tools like Apache Nifi.

The multiple pipelines enabled data specific transformation rules and validation checks, thereby allowing for more flexibility and granularity in the data processing.

Additionally, to automate this process, the team used UC4, a commercial automation software to schedule and orchestrate the movement of data from an on premise system to the cloud. AWS Batch validation jobs (data quality assurances, data duplication checks, and data completeness checks) were now run to make sure the transformed data complied with the required standards and were prepared for loading into the Snowflake data warehouse.

Once the data is loaded into Snowflake, it is accessed and used by business users for reporting, analysis, and decision making.

In addition to migrating to Snowflake, the team at Factspan also implemented a purging mechanism (a process used to regularly remove or archive old or unnecessary data from the data warehouse) to manage the stored data in the data warehouse.

Business Impact

• Reduced operational costs and enabled real-time data transfer, visibility, and automatic upgrades
• Reduced the run-time of batch data by 15-20%
• Optimized the parallel processing capability from 40 to 400+ processes.
• Easy business scalability and application elasticity even during spikes & increased customer demands