Big Tech

Equifax Slashes Database Support Tickets by Moving Billing to Oracle Autonomous Database on AWS

The credit bureau migrated its massive billing database from Amazon RDS to Oracle's Autonomous AI Database, eliminating the need for query tuning and cutting support overhead dramatically.

2 min read
Equifax cuts database firefighting after moving billing to Oracle on AWS

The nature of multicloud strategy has shifted. Rather than enterprises choosing which provider hosts which workload, database vendors are now bringing their services directly into the cloud platforms where customers already operate. This shift gains additional momentum as artificial intelligence projects expand the business case for such moves beyond routine back-office functions.

Equifax Inc., the credit reporting firm, exemplifies this trend. The company has transitioned the vast majority of its on-premises data centers to cloud environments and distributes its operations across Amazon Web Services Inc., Google LLC's Google Cloud, and Oracle Corp.'s cloud infrastructure, according to Avinash Singh, vice president of technology services at Equifax.

Eliminating the database tuning burden

Equifax's worldwide billing infrastructure depends on Oracle's billing and revenue management platform, which processes and rates billions of transactions monthly. Previously, the underlying database operated on Amazon Relational Database Service for Oracle, but the system approached a 65-terabyte storage limit. Oracle Autonomous AI Database on AWS, powered by Exadata infrastructure, offers petabyte-scale capacity within the same cloud environment as enterprise adoption accelerates. Since the broader application architecture remained on AWS, only the database layer required replacement, Singh explained.

Their performance optimizer is so advanced that we don't need to tune a single query. Just think about that. Fifteen to 20 tickets per day turns into one to two tickets per month.

Avinash Singh, vice president of technology services at Equifax

The migration itself proved straightforward from a development standpoint. "There is zero code writing because it is just about migrating the data from one database to another," Singh said. "We achieved this migration without any downtime."

Performance gains after go-live

Following the April launch, performance improvements materialized immediately. The process of collecting database statistics across nearly 100 terabytes of data compressed from three to four days down to three to four hours. Following some configuration adjustments, batch operations that previously handled 100 to 200 rows per transaction now manage 15,000 to 30,000 rows, according to Singh.

https://www.youtube.com/embed/6WuK0oCGvJ8?feature=oembed

Our rating improved almost 200% to 300%. We are able to close our books way before what we have been closing.

Avinash Singh, vice president of technology services at Equifax

Source: SiliconANGLE · Reporting supplemented by The Silicon Ledger staff.