Advanced Data Analytics on ClearPath Forward:
A Case Study

What do you do when your organization is processing more data everyday – and changing its operational structure at the same time – but the systems you have in place to extract and analyze this data can’t keep pace?

That’s the dilemma we recently helped one of our clients address.

The trouble started in the organization’s existing analytical database. Though its ability to handle around two gigabytes of data at a time, access four months of stored data, and process roughly two million transactions had been sufficient, the client’s changing operational structure and evolving business requirements made this current model insufficient.

With a plan in place to spread its operations across defined geographic zones, the client suddenly needed to extract, process, and perform predictive analytics on large amounts of data, so it could better combat fraud. Role-based access was also an important requirement in the area of data security. These changes put immense stress on the organization’s existing database environment, emphasizing issues around the shareability of information, as well as the speed and efficiency with which it could process these growing volumes of data.

Given these issues, the client put a new plan in place: Improve its data storage and retrieval capabilities using a database that’s able to house large volumes of information.

As part of this effort, the client wanted to allow its users to tap directly into data around the clock, and perform intricate, robust analysis – without affecting performance for users in other locations. And they needed to be able to do this using multiple data mining, statistical, and visualization techniques.

The Unisys Solution

After listening to the client’s goals and concerns, we built a Proof of Concept (PoC) solution that illustrated how to pull together several data transfer and visualization tools, while continuing to leverage the company’s ClearPath Forward™ system.

To that end, we recommended installing the Red Hat® Enterprise Linux® operating system and the Hortonworks Data Platform® for Apache™ Hadoop® on the client’s ClearPath Forward system. Together, these technologies would form the core Big Data platform underpinning the rest of the solution.

In the area of data analysis and visualization, we selected the RStudio integrated development environment and Pentaho business intelligence software. In the PoC deployment, RStudio facilitated predictive analytics by performing the statistical modelling required to spot fraudulent activity in the client’s environment. Meanwhile, the Pentaho software was used to build a data pipeline between the company’s ClearPath® MCP environment and Hadoop data lake, while also helping end users visualize data in reports and dashboards.

To facilitate the Extract, Transform, Load (ETL) process from the client’s Enterprise Database Server for ClearPath MCP (DMSII) database to the Hadoop environment, the PoC solution created a data pipeline using the Relational Database Server for ClearPath MCP JDBC resource adapter, Apache Sqoop™ tool, and Pentaho Data Integrator (PDI).

With these technologies in place, the PoC project helped to show how the client could efficiently and securely make over seven years of financial and credit card data directly available to its user population. Accessing this information from a scalable foundation that can extract and process data from different sources in parallel meant the organization would no longer handicapped by growing data volumes.

And with the ability to process large amounts of data quickly, cost effectively, and with a minimal impact on internal resources, the organization should be able to get more value from this data than it previously could. This way, the client can make decisions with the speed, accuracy, and certainty required to confidently enact new operational strategies. Moreover, the solution is expected to help the client increase revenue by enabling it to effectively detect – and proactively prevent – fraud.

Share this article: