As we know the PureData System for Operational Analytics can deliver actionable insights concurrently to more than 1,000 business operations to support real-time decision making. Operational warehouse systems are used for fraud detection during credit card processing, to deliver customer insights to call center operations, and track and predict real-time changes in supply and demand for energy and utilities.
Scenario: Fraud Detection during credit card processing
The PureData System for Operational Analytics can handle both short, operational and longer, complex, analytical queries. For example operational queries can be to the tune of 1500 queries per second or over 5 million queries per hour. The queries to check if the credit card usage is fraudulent or not (short, operational query) needs credit card user's profile and history data which cannot be obtained from an OLTP system and hence a data warehouse system is required. However remember high number of credit card requests and transactions are being performed simultaneously (short, operational). Analytical queries analyzing large amounts of data (longer, analytical workload) is used to determine patterns of credit card fraud and these on the other hand have larger run-times, with several executions per minute.
Figure 1: A snapshot from the management system console:
PureData System for Operational Analytics is able to handle concurrent usage of operational and analytic workloads - managed by the underlying InfoSphere Warehouse - DB2 workload manager (more CPU shares provided to the operational queries), and ensures the operational queries are not affected by the analytical queries and thus able to deal with such mixed workloads.
Figure 2: A snapshot from the performance monitoring console - checkout the number of transactions per minute near 50K:
Watch the demo and more at the IBM IOD - PureData System for Operational Analytics Demo Station 16 , Real-Time Operational Warehousing - Station 9