Verisk: Migrating to Amazon Aurora with AWS Data Lab Case Study

Verisk chose to collaborate with AWS Data Lab for expert guidance on these migration challenges. In April 2019, Verisk participated in its first AWS Data Lab to focus on the migration of two databases that included tables that were partitioned with billions of rows and required complex transformations for LOB datatypes. Using AWS Database Migration Service (DMS) and AWS Schema Conversion Tool (SCT), AWS Data Lab helped the Verisk team architect and prove out a migration path to move data from two on-premises databases to Amazon Aurora PostgreSQL, capture change data, and integrate its applications with the new target databases. Verisk chose Amazon Aurora PostgreSQL because of its high availability, backup and restore capabilities, cluster cache management, fast cloning, and replication functionality. The Verisk team finished the week with greater technical skills and a proven, repeatable process it could use to migrate Verisk’s remaining data to AWS.

Several months later, a second and third business unit at Verisk uncovered additional complex migration challenges. In December 2019, Verisk chose to re-engage AWS Data Lab to migrate two of its most complex schemas from its legacy database to Amazon Aurora PostgreSQL using AWS SCT, refactor and modernize its existing PL/SQL application code to Java, and refactor 14 legacy procedures into a single procedure. “Quality expertise was provided for us to break through processes vastly quicker than self-research. Many tips were provided around best practices and possibilities that we would have not realized we needed to know for months. The AWS Data Lab format was very good, allowing us to take use cases from our office to work on and then provided artifacts to bring back as working models,” shared a Verisk builder who participated in the AWS Data Lab.

Soon to follow in January 2020, Verisk’s third team joined the AWS Data Lab for guidance migrating from its legacy spatial database to AWS. The AWS Data Lab helped Verisk choose the right AWS database option based on its business requirements, which was Amazon Aurora PostgreSQL. In only four days, the Verisk team used AWS DMS and SCT to successfully convert a subset of its spatial schema to PostgreSQL with PostGIS, migrate 85 spatial and 135 non-spatial tables to PostgreSQL, and run initial functionality tests on its downstream application after integrating it with the new target database. The AWS Data Lab provided Verisk with access to an AWS DMS feature for migrating spatial data to PostgreSQL, which at the time was not publically available. The feature was released later in Q1 2020. Another Verisk builder noted, “As a Database Administrator at Verisk working on the data migration, I am miles ahead of where I was prior to working with the AWS Data Lab. I have more confidence in being able to successfully migrate our legacy database to Aurora PostgreSQL and have a better understanding of what products are available to us. I couldn't have asked for a better experience.”