alphaspirit - Fotolia
As monolithic applications give way to microservices and API-driven modules, integration of data grows increasingly complex. In this discussion, Liaison Technologies' Madhukar Kumar and Patrick Adamiak talk about how data management can help IT unlock the power of a data-centric business approach. They explain the concept of exposing data as an API to break down barriers between data silos. Doing so allows the business to tap into information that it can combine to make better decisions. Liaison develops integration products through its data platform as a service.
They also talk about the need for application modularity that is extensible and customizable, and the issue of data veracity -- in other words, making sure that data is clean, consistent and harmonized before use. This step helps companies work with multiple types of data, which might be as varied as feeds from an oil platform and customer experiences on social media. Regardless, the purpose is to gain business insights and glean ideas from information that were not previously extracted. People are excited by the prospect of putting together seemingly random sources of data and getting from those sources a new idea for running their business.
One challenge, they point out, is the shortage of in-house expertise to plan and carry out data-centric projects at all but the largest of corporate enterprises. Adamiak, vice president of product marketing at Liaison, says it's a complicated problem and that's why businesses contract with vendors to provide the hardware and data management for such projects.
Another key for businesses is data integration. Using an example of Salesforce sales and customer management with NetSuite ERP, even if a change made in one system automatically updates the other, that's not good enough, explains Kumar, vice president of product management at Liaison. If, for example, you want to find how many customers you currently have, how many came to your website and how many customers you might have lost, the data is sitting in two different places and must be combined. Doing that requires several steps, including extract, deduplication and matching up corresponding records, Kumar says.