The concept of massively parallel processing (MPP) used to be the sole province of supercomputers - machines used in scientific and government installations to measure data-intensive problems, such as weather forecasting or defense analytics. Today, almost any business can harness the power of MPP, especially when it comes to analyzing data within its own applications.
Companies such as Dataupia have taken advantage of low-cost commodity hardware to build powerful but expandable appliances that can offload highly intensive queries from traditional database applications. And the capability applies to any application that relies on a database – whether a traditional Oracle or SQL server application, a customer-relationship management application or business intelligence tools.
MPP brings a level of speed and flexibility previously uncommon. "One of our customers had a typical query that took 7-1/2 days to run," says Samantha Stone, vice president of marketing at Dataupia. "The result of the query had value, but only to those who could wait that long." Most people can't wait that long; the business opportunity afforded by the information has probably passed.
Using the MPP technology in Dataupia's appliance for data warehouses, the query now takes three hours, Stone says. By taking an aggregated view of the data, the appliance can respond to queries in just a few seconds. "The company has opened up a whole new set of users who can benefit from that type of response – C-level executives looking at dashboards, line management executives looking for trends and patterns, and other people within the corporation who have never been given access to historical information before." Stone cites the example of customer service representatives who can handle queries based on customers' entire purchase history rather than the previous 30 days.
In the architecture of its appliances, Dataupia takes further advantage of MPP design by offering more flexibility, according to CTO John O'Brien. Previously, companies wanting a 10-terabyte data warehouse solution needed to buy an appliance with that capacity. When the company grew, it would have to swap the first appliance for a larger one. "Our appliances act as part of an array. You can build a system based on what you need today," O'Brien says. "As your business grows, you can add more capacity easily."
This means that even a midsize company running an SQL server database can take advantage of MPP to add insight into patterns and trends, gaining a competitive advantage against larger companies with bigger databases. With this expandability, O'Brien says, "Any company can start out relatively small and get into a machine that grows as its business grows."
 |
 |
|
« Previous Article Tackling the Challenge of Real-Time Data Flows |
