American Express has converted data on more than 60,000 projects from its legacy project portfolio management (PPM) system to a more standardised system in just five days.
Speaking at CA Technologies’ CA World conference in Las Vegas, American Express explained how it managed to perform the large-scale data conversion with minimal business disruption. Its aim was to move project data from a legacy PPM system, developed in-house for 12,000 end users, to the Clarity PPM tool from CA.
“Back in 2007, we had a legacy PPM tool that we built up and managed ourselves for about seven years.
“A lot of the transformational strategy was to standardise our technologies and automate. The other [legacy] tool set was very manual and involved a lot of work to manage and keep it up,” said Brent Beenau, Director-Technical of Systems Development Life Cycle (SDLC) Tools Delivery at American Express.
According to Beenau, the Clarity PPM tool allowed the seamless transition of historical data, and the integration with the system of records.
The main challenge that American Express faced with the large-scale data conversion was the speed at which the project could be carried out, in order to minimise disruption for its users. Initial testing showed that it would take to 40 days to complete the conversion.
“Working with CA, they continued to refine the process until the desired result [five days] was achieved,” said Chiranjib Misra, Director-Technical Delivery at American Express.
To this end, American Express updated its infrastructure by developing JVM Parameters on its application servers, upgraded its Oracle database from 10.1 to 10.2.0.4, and set the CPU speed for Oracle. The legacy system had been built up on a SQL server.
Meanwhile, on the Clarity side, the credit card company upgraded from CA PPM 8.1 FP01 to FP04, used SQL Trace and worked with CA on hot fixes.
However, before converting the data, which included 12,000 resources linked to the 60,000-plus active, pending and closed projects in 2009, American Express decided to clean it up first, to reduce the level of rejection.
Misra also said: “We didn’t want to start with a system with data that had lost its meaning over time.”
The company cleaned up the data in a number of ways. These included closing open master projects with no “active children” and closing projects that had not seen activity for more than 90 days.
“When they’re closed, they’re static project information that you can move over and they won’t change,” said Beenau. In doing this, the company found that 30,000 of the 60,000 projects were or could be closed.
Other clean-up activity involved updating missing email addresses, cleaning up vendor names and deactivating resource IDs that have not recently seen activity.
American Express also developed a Clarity data dictionary to manage the configuration of information, which Beenau said the company still uses to this day.
“A Clarity data dictionary was created to manage the configurable data elements. Fifty-five percent of data elements required customisation or translation from the legacy PPM to Clarity,” he said.
He also said that by mapping the data, he was able to discover that 25 percent of the legacy PPM data elements were not required for conversion.
In addition, Beenau and Misra emphasised the importance of people in the carrying out the data conversion.
“The people side was really important because our implementation was global. We had to think – how do we convey the message of change? We worked with leadership teams, at board level, to bring them on board to help manage the change,” said Beenau.
“You’ve really got to think about the end users.”
To support the end users, the company set up a command centre during the implementation of the new system. This provided end users with access to online reference tools, for example.
The centre was in operation one month before the implementation and ran for 90 days until America Express saw that the system had stabilised, at which point the command centre was integrated into the company’s traditional service desk.
Beenau added that the company was keen to collaborate CA “early on in the process” because it is “so much easier if you have roadblocks”.