Cloud Migration Cloud Ops

Equifax moves billing database to Oracle Autonomous AI Database on AWS

Equifax has moved the database supporting its global billing system from Amazon Relational Database Service (RDS) for Oracle to Oracle Autonomous AI Database on AWS, reporting faster processing and a sharp reduction in database support tickets. The migration changed the database service while leaving the application stack on AWS, according to Avinash Singh, Equifax vice president of technology services.

Singh described the move at Oracle AI World. Equifax’s billing and revenue management system uses Oracle software to rate and invoice billions of transactions each month. Its previous Oracle database deployment on Amazon RDS reached a 65-terabyte limit, prompting the company to move to Oracle’s service on AWS, which runs on Exadata hardware and can accommodate petabytes of data.

Database migration without an application rewrite

Equifax kept the billing application on AWS rather than moving the full system to another cloud. Singh said the migration required transferring data between databases, not rewriting application code, and was completed without downtime. The system went live in April.

The approach illustrates a form of multicloud deployment in which a company can use services from multiple providers while retaining workloads in the cloud where they already run. Equifax has moved nearly all of its on-premises data centers to the cloud and operates workloads across Amazon Web Services, Google Cloud and Oracle Cloud Infrastructure, Singh said.

For this billing workload, the change addressed a database capacity constraint without relocating the surrounding application. The new service also reduced the time required for a database maintenance task: collecting statistics across nearly 100 terabytes of data fell from three to four days to three to four hours, according to Singh.

Reported gains in processing and support

Equifax also adjusted the system’s configuration after migration. Singh said batch jobs that previously processed 100 to 200 rows per transaction now handle between 15,000 and 30,000. He said the company’s rating performance improved by almost 200% to 300%, and that it can close its books earlier than before. The remarks did not specify a baseline or a measurement methodology for the performance comparison.

Operational support needs also changed, according to Singh. He contrasted 15 to 20 database tickets per day before the migration with one to two tickets per month afterward, attributing the reduction in part to the database service’s performance optimizer. He said Equifax no longer needs to tune individual queries. The figures are his account of the system’s results.

The case centers on a production billing database rather than a broad replacement of Equifax’s cloud strategy. The company retained AWS for the application stack and adopted Oracle’s database service within that environment, avoiding downtime during the transition as reported by Singh. Its experience highlights how a database service change can address capacity and operational issues while preserving the location of an application workload.

Equifax’s deployment also shows that multicloud use need not mean moving each application between providers. In this instance, the company’s wider environment spans three cloud providers, while the billing migration paired Oracle database technology with an application that remained on AWS. The reported improvements followed the service change and subsequent configuration tuning.