Hewlett-Packard's business intelligence unit Monday announced a new release of its Neoview data warehouse that promises to solve one of the most vexing problems of operational business intelligence -- balancing a high-volume of short queries, like those from customer service representatives focused on a single customer, with larger, analytical jobs such as a comprehensive analysis of company operations.
Neoview 2.3, the second largest update to the data warehouse since it began shipping in October 2006, adds several features created to allow it to more efficiency handle such mixed workloads, HP said.
Neoview, which integrates data warehousing hardware, software and services, is being used in the US by Wal-Mart and HP itself.
"We view operational BI as a competitive weapon that companies can use to bring better differentiators to themselves in this increasingly commoditized world," noted Vish Mulchand, director of Neoview product marketing. "These kinds of capabilities are crucial to giving customers the insight they are looking for."
The new release, which is now available, includes Adaptive Segmentation technology, which boosts performance by automatically matching the appropriate computing resources to different sizes and types of queries, added Greg Battas, Neoview's chief architect. While large massively parallel systems have traditional been very strong at solving the complex queries, they are sometimes hard-pressed to handle a large number of small queries at the same time, he noted.
"Our system now will virtually segment itself up it," he said. "It might decide to only use a small slice of the machine for certain queries. We're able to handle much higher concurrency rates by virtually slicing the machine up."
Another new feature, dubbed HP Skewbuster, acts as a sort of traffic cop that prevents bottlenecks -- such as one that could arise from skewed data that appears to associate one customer with all customer purchases because of customer ID numbers being were incorrectly entered.
Neoview 2.3 also allows the database to "ingest" data without it having to be physically loaded into the warehouse so that companies can be assured they are analyzing the most up to date information when performing tasks like identifying fraudulent transactions in real-time so an action can be taken.
Finally, new workload management software in this release constantly monitors how busy the system is, how many users are accessing it, how many resources are free and other factors to ensure that jobs are queued up to ensure that the most pressing queries - such as those associated with an employee speaking to a customer in a call center - are prioritized over other jobs that could perhaps wait a few minutes before starting.