Challenge
A financial industry leader, famous for its credit cards, Samuel L. Jackson- starred “What’s in your wallet?” commercials, Capital One’s applications generate huge amounts of data. Mining of which has become of paramount importance for making critical business decisions. Capital One was storing this data in multiple warehouses and databases and using traditional batch analysis for decision making. They needed to quickly process and analyze their data to make faster decisions in near real-time. Capital One sought quickly to build a new, real-time reporting and analytics platform. “It’s a challenge that the business is going through from thinking about batch analysis to a near-real-time platform,” said Javed Roshan, Director of Data Services at Capital One.
Solution
The stringent service level agreement (SLA) requirement and the need to quickly process large amounts of data made DataStax Enterprise a natural choice for Capital One. Capital One attained a distributed environment out of the box with multi-datacenter replication. This environment is also extremely flexible and able to handle incredibly fast writes, which Capital One required, and has tunable consistency for different types of queries. Furthermore, reliable scalability makes it easy to grow data clusters and support new features, and DSE Search allows for flexible and customized querying. “Along with scalability, availability, and real-time capabilities, DSE Search provides many options to query data like range queries, regular expression, and non-equality,” said Roshan. “These features are very tempting to use.”
Results
Consolidating this into far fewer rows (11.5 million), all handled on a six-node cluster at the lightning-fast rate of 21,000 transactions per second and meeting Capital One’s 99.99% uptime requirement.
The new platform kick-started Capital One’s big data intelligence by quickly and easily incorporating a year worth of data – 450 million rows – from a legacy Oracle database. Consolidating this into far fewer rows (11.5 million), all handled on a six-node cluster at the lightning-fast rate of 21,000 transactions per second and meeting Capital One’s 99.99% uptime requirement. “We pulled in one year of data from Oracle and once we got it into DataStax Enterprise, built on the best distribution of Apache Cassandra, it was a smooth ride and it was processing at a very high rate,” said Mukram Aziz, Senior Manager of Data Services at Capital One.
