CASE STUDY - Keeping Data Actionable with Changed Data Capture
By Mitchell Light, Senior Technical Analyst, Syncsort Incorporated
Business intelligence cannot be collected from stale data. The explosive growth of information, coupled with increasing demands for customer service, has left many companies looking for more efficient ways to keep data actionable. Refreshing and maintaining customer databases using changed data capture (CDC), the process of isolating changed records, is one such method. Using DMExpress for CDC, a major provider of telecommunication services saw a 92 percent improvement in its nightly customer care process.
The database and systems administrators at this nationwide telecommunication company recognized the need to upgrade their internal systems. One of their goals was migration of vast customer care databases from mainframe to UNIX with subsequent implementation of another vendor’s management, monitoring, and automation software infrastructure.
In the company’s mainframe environment, refreshes had been executed every night. Now on UNIX, the arduous PS/SQL process, plus the massive amounts of nightly loads, became unwieldy and prevented implementation of the other vendor’s infrastructure. CPU and memory bottlenecks, caused by the enormous refreshes, needed to be unclogged.
The administrators realized that instead of loading upwards of 57 million records per night, which the other vendor’s refresh required, they needed to load only a fraction of those records—those that had changed or were new. They set off to implement a method to isolate changed and new records, and they discovered that the CDC/delta processing is among the most resource-intensive data processing functions.
With processing still on the mainframe, the administrators unsuccessfully sought a way to perform efficient CDC using the tools they had in house. These efforts continued throughout the mainframe-to-UNIX migration process. Finally, they turned to Syncsort Inc., which offered to demonstrate the CDC capabilities of DMExpress, which is capable of fast and sophisticated joins, sorts, merges, and summarizations.
It took only one demonstration for the telecom company to be convinced that DMExpress provided the solution they sought. Syncsort’s proof of concept was performed on the company’s actual data and equipment and was immediately adopted. The proof of concept was carried out on the company’s existing databases, running on a second vendor’s servers with four dual core processors in their operating system.
By exploiting its advanced data management (ADM) functions, as well as the first vendor’s source and target features, DMExpress was easily inserted into the customer care nightly job flow, where it now swiftly identifies new and changed records and then neatly updates the customer care databases each night. The total savings in elapsed time is an astounding 92 percent—reduced from four hours to 20 minutes. And, more importantly, the process has freed CPU and memory so that the company has been able to implement the first vendor’s infrastructure and complete its year-long conversions.
Through Syncsort’s free proof-of-concept program, this telecommunication company learned that DMExpress’s CDC is the fastest and most efficient in the industry. Their database and systems administrators are happy with the results: freed CPU and memory, and significant savings in elapsed time for their overnight jobs. These improvements have enabled the company to provide improved customer service. Plus, as one DBA asserted, they are already discovering new, unexpected ways to make beneficial use of DMExpress.