Cost considerations and easy-to-understand packaging recently led a $3.2 billion financial services holding company to choose open source data integration tools
Headquartered in Ithaca, N.Y., Tompkins Financial Corp. owns three banks with more than 40 branch offices, an insurance business and several wealth management companies. The publicly traded financial services organization is subject to many federal reporting regulations, including Form 10-K, an annual report required by the U.S. Securities and Exchange Commission that provides a comprehensive summary of company performance.
To fulfill its various reporting requirements, Tompkins regularly pulls information from ERP systems, general ledgers and various other information sources spread throughout its businesses. But upon joining the company two and-a-half years ago as vice president of IT, Glenn Cobb saw that his new firm desperately needed to modernize its data integration, information gathering and business intelligence (BI) reporting techniques.
Information about Tompkins’ individual customers resided in multiple systems and the core banking systems generated reports electronically -- but just barely, according to Cobb.
“It was just sort of a mess,” Cobb said. “[The reports] were something you would normally print out.”
Compounding the problem was the fact that several older members of Tomkins IT staff – people familiar with Assembly – had been with the company for 20-30 years and were gradually beginning to retire one by one.
“Nobody knew how this stuff worked,” Cobb said. “So, I went out and started looking at different [extract, transform, load (ETL)] products. ETL was part of my previous jobs and I knew the market a little bit.”
A data integration tools evaluation
Cobb began the data integration tools product evaluation process by looking at IBM InfoSphere DataStage. IBM says DataStage is an ETL-based offering that supports the collection, integration and transformation of large data volumes of data in “real-time” as well as data received on a scheduled basis over time.
“I was [previously] certified in DataStage, so I had a good understanding of object-based modeling and [ETL] functions,” Cobb said. “But they were too expensive. They were $100,000 plus just for a couple of seats. It’s a great product but for a midsized community bank [that is] hard to justify.”
Cobb and his team then looked at the Microsoft SQL Server Integration Services platform as well as data integration tools from Informatica, but they didn’t quite fit the bill.
“I went through all of these products and tried to identify what would be the most flexible, economical and able to fit our business needs from a usability, cost and ease of adoption standpoint,” Cobb explained.
Tompkins ultimately decided to go with open source data integration and ETL tools from Talend. The company also purchased Talend data quality tools to ensure the accuracy of the information going into its reports.
Cobb said the decision to go with Talend was largely driven by cost considerations, but he was impressed by the fact that he could easily test out the Talend software on a shareware platform. He also liked Talend’s approach to pricing and packaging its applications.
“The problem with the IBM product is that there are four or five components. If you want to do data quality, that is another module. If you want to just do data transformation, that is a module. If you want parallel versus a single channel, that is a different scenario. It’s a very costly and you had to have a lot more installs and stuff,” Cobb explained. “The thing I like about Talend is that it’s a single install and one product it does everything.”
Implementing Talend data integration tools – a tricky process
Tompkins has been running Talend for the past year and using the technology to automatically convert data into records that are ideal for BI and federal reporting requirements. Cobb said the manual processes that Tompkins formerly employed to complete the 10-K report took several hours per month, but that time has now been cut down to minutes.
That said, Cobb cautioned that the task of installing Talend software turned out to be more difficult than he initially expected.
“You have to install a SQL database, whether it’s MySQL or [Microsoft SQL] because [Talend] needs a place to store the log files,” he said. “You have to make sure that some other services are installed and running and then you have to install the major components and configure them. The installation guides are 40-50 pages long.”
The IT veteran cautioned that companies who choose Talend should make sure they have skilled people on staff with experience doing similar implementations. He said it’s also a good idea to work closely with Talend support staff throughout the process.
Yves de Montcheuil, Talend’s vice president of marketing, explained that the core Talend products are open source, but the company also offers proprietary “value-added” capabilities that are proprietary in nature.
He added that the Talend product portfolio reflects a growing trend in the marketplace toward consolidation of all data management activities, including data quality, data integration tools and master data management (MDM).
“Those projects need to be together as global data management discipline,” he said.