A pressing need to leverage massive amounts of Big Data in the name of business agility has put shipping giant Federal Express on the path toward re-engineering most of its enterprise applications.
According to FedEx CIO Rob Carter, that need to analyze events in real time has resulted in an effort to “radically” decompose monolithic vertical applications into sets of core granular services, which the company will then mash into any number of analytics applications. The ultimate goal: a matrix of IT services that functions with the speed and flexibility of a brain, freeing FedEx from a system dependent on files strewn across any number of databases kept on disk storage systems too slow to support advanced, real-time analytic applications.
“The line between memory and storage is blurring,” he said. As memory becomes less volatile, it becomes more feasible to process everything in memory versus making calls to disk. Carter envisions the next generation of the company’s analytics applications as capable of accessing data stored in memory, which in turn could quicken the supply chain’s reaction times—for example, automatically rerouting shipments in the event of rioting in Spain or a flood in Thailand.
Carter said that FedEx already collects massive amounts of Big Data using the EMC Greenplum platform and database appliances from Teradata. As processors evolve, it’s becoming more and more apparent that databases will be relegated to managing large amounts of archival storage housed on disks; predictive analytics, meanwhile, will end up driven much more by in-memory computing.
Much of FedEx’s data comes from sensors, including a new SensaWare service that involves placing active sensors inside boxes containing high-value merchandise. Rather than wait to be scanned, these sensors actively send out telemetry data covering everything from the package’s traveling speed to whether any light has penetrated its packaging (the latter would suggest some sort of issue has occurred).
While these sensors are too costly to place in every FedEx shipment, Carter noted that, as the cost of sensor technology drops, it could become more practical to broadly apply the technology. The real challenge, he added, is making sure the backend IT infrastructure is in place to handle all the associated data processing and storage.
Longer term, Carter expects to make all that data more readily available via application programming interfaces (APIs) that would allow customers to apply their own analytics applications against the data. Within FedEx, the company’s applications are all connected to a common message bus that allows applications to publish and subscribe to any source of relevant data.
Those capabilities will prove critical as modern enterprise applications evolve to broker data between services. In fact, it’s a core capability embedded in Windows 8. “This is one reason we’re actually excited about a new Windows platform in a very long time,” Carter said. Because the operating system now manages the brokering between sources of data, the complexity of building composite applications that access multiple data should be significantly reduced.
While there are still major challenges when it comes to delivering actionable intelligence in real time, Carter believes that FedEx can benefit from advanced analytics technologies more easily than some other businesses because its business processes are more structured; it can leverage historical data and new information in the context of a predictive-analytics application, for example, to plot a new shipping route.
In addition, Carter says that FedEx uses analytics to actively monitor social networks and video feeds, identifying customer service issues in a more proactive way. FedEx has also set up collaboration spaces through which customers can share information provided by FedEx with customers, partners and suppliers.
While next-generation analytics applications are helping drive business innovation, Carter believes that businesses must modernize their backend IT infrastructure in order to take full advantage of advanced analytics applications. That’s not something that most IT organizations are going to achieve overnight. But it’s pretty clear that advanced analytics are the tipping point of a transformation that extends from sensors on packages all the way to the data center.
Image: Eugene Berman/Shutterstock.com & TheNextCorner