Monitoring system reveals when customized SAP and Oracle application operations, common in packaged software implementations, result in performance hits.
A monitoring system from Cast Software is geared to measure the performance of enterprise resource planning applications. It's also aimed at revealing when those customer customizations, common in packaged software implementations, result in performance hits.
SAP's ERP applications, Oracle's PeopleSoft HR applications, and Oracle Siebel customer relationship management apps all require intensive configuration by customers, with no two customers doing each setting exactly the same way. Then it's common for customers to customize, add, or change code to make the application fit better into business operations. As a result, there's little shared knowledge on how to get the applications to best perform, and even less on when the customizations are leading to performance degradation.
"SAP, PeopleSoft, and Siebel are among the most difficult environments to manage," said Vincent Delaroche, Cast founder and CEO, in announcing enhancements to the Application Intelligence Platform at the end of May.
"We work on looking at very complicated enterprise systems and understanding what they do," said Jitendra Subramanyam, director of Cast strategy and research, in an interview.
Enterprise programmers are frequently called on to customize SAP or Oracle applications once they are installed. Outside consultants, hired to do the implementation, perform customizations in order to meet customer expectations. Performance issues arise when the customizations, such as calls to databases, are done with a poor understanding of how the application operates.
"A lot of these database calls are not made correctly. They lead to erratic performance issues... Weird issues happen. The problem lies in the breadth of knowledge of the person working on an application," Subramanyam said.
Cast's Application Intelligence Platform has been upgraded to provide better assessment of SAP, Siebel, and PeopleSoft applications. It parses an installation to analyze the source code of the application and build a model of its current configuration. Then it runs "sophisticated pattern-checking algorithms against the model to see it the user is following known best practices" for adding to or customizing the system.
Subramanyam said Cast has 100 pattern-checking algorithms and "a huge database of known hotspots" in applications that lead to performance degradation. The result of its analysis is "a set of measurements of quality, size, and complexity. We quantify those in a precise way."
Dashboards can be built to focus on particular performance metrics that are important to the business, and consultants, enterprise developers, and software architects can all check the feedback on what they should do to improve performance. Operations managers can also use the analysis as a way to pinpoint where system resources are being swallowed up and use the information in scheduling jobs until problems can be resolved.
Subramanyam said IBM Global Services, Accenture, AtosOrigin, CSC, and CapGemini have all made use of the Cast analytics in their consulting engagements.
"We can very efficiently decide whether it's an application problem or a network problem. We can help in root cause analysis," he added.
Part of what Cast is attempting to do is fill a gap between the function test analysis done by code developers and testers and actual application performance, as seen by operations managers. Function tests show whether code will fulfill the function set by system requirements, but they don't show how well those functions will perform with other parts of the application and its related middleware and database systems. Once in a real-world setting, code often doesn't run as planned. And when it's customized further by outside hands, say a systems integrator or consultant, new burrs are added to the finicky application machinery that can have unpredictable results, Subramanyam said.
The Application Intelligence Platform has been sold since 1997 and Subramanyam said it has an ability to parse more application code than any other analysis system. Cast has 250 customers worldwide, with a majority in Europe. A Cast system typically costs $450,000 or more, depending on the number and complexity of applications analyzed.
Cast was founded by Delaroche in Paris in 1990, and went public as the first initial public offering on the Euronext stock exchange in 1999. Delaroche was a major contributor to the French book Software Quality in 1992, and founded the publication, IT Expert, a leading French publication on IT expertise. He moved to New York City in 2005, and Cast maintains headquarters in both Paris and New York.
Join us for a roundup of the top stories on InformationWeek.com for the week of December 14, 2014. Be here for the show and for the incredible Friday Afternoon Conversation that runs beside the program.