MetLife improves efficiency with Tibco Data Virtualization

The problem

With millions of customers in dozens of countries, MetLife collects copious amounts of data from many disparate sources.

Until 2017, however, MetLife was using a system it had developed in house to ingest and integrate all that data coming into MetLife in all its different forms.

All the work to standardize the data was manual, including avoiding data duplication, and it took two to three weeks for MetLife to integrate data into its data warehouse and transform it into data sets that could then be used for analysis, according to Paul Hammond, data virtualization architect at MetLife.

Going back to 2014 or 2015, we realized that data integration was a big problem. We had grown as an organization. We weren't quite at the big data stage yet, but we knew we needed to get a better way of handling data.

In fact, it was Hammond in his role as senior data warehouse architect from 2010 to 2015 who was responsible for the management and architecture of MetLife's global data warehouse.

The data warehouse was already in place when Hammond began managing it, and though the legacy system was starting to fail to meet growing demand, it was too cumbersome to modernize given all the other technology already connected to the data warehouse and all the users accustomed to the incumbent system.

Eventually, however, MetLife came to the conclusion that it needed to revamp its data integration technology.

"Going back to 2014 or 2015, we realized that data integration was a big problem," Hammond said. "We had grown as an organization. We weren't quite at the big data stage yet, but we knew we needed to get a better way of handling data."

The solution

Finally, in 2017, MetLife took steps to reduce its data delivery cycle and ease the manual burden on its data management staff. And despite fears that modernization would be cumbersome and disrupt long-established workflows, the organization found a platform and an approach to data integration that simplified the modernization process.

The organization decided to take a data virtualization approach to its data management, implementing a data virtualization platform to automate certain processes and aggregate its data to create a single view. In April 2017, it decided to use Cisco's Data Virtualization platform, which was acquired by Tibco by the time MetLife was done testing the platform and able to implement it late in 2017.

MetLife tested the data virtualization platform across multiple departments before choosing it, and Hammond immediately recognized its benefits.

"I came out of the proof of concept going, 'This is a really cool tool -- we can use this everywhere,'" he said. "With it, we didn't have to do staging in our data warehouse; we didn't have to mess with files anymore and could use this data tool to solve the majority of our problems.

"It's what we had been looking for," he continued. "My whole thing was how to make it easier for people to get the data."

MetLife's challenges, meanwhile, were not uncommon, according to Mark Palmer, Tibco's general manager and senior vice president of analytics, data science and data products.

"It's the reason we exist," he said. "From a data integration perspective, data virtualization elegantly virtualizes access to all this data instead of dealing with a rat's nest of data moving around. MetLife's is a typical story that we hear all the time."

The result

According to Hammond, MetLife's typical data integration process took as long as three weeks before the company started using Tibco Data Virtualization.

The old process -- all done manually -- meant adding fields to the data warehouse, but first taking a few days to examine all the downstream ramifications of adding new fields. Then it took a few more days to modify SQL Server Integration Services packages, and finally yet another few days to process a data-modeling request.

"Basically, we were looking at a two- to three-week build cycle," Hammond said.

Now, with Tibco Data Virtualization, the build cycle takes less than three hours, he continued. And MetLife could get certain results in real time with Tibco's platform, but hasn't yet taken advantage of those capabilities.

"We haven't realized the real-time access to data, but we know it's there," Hammond said. "The real benefit that we've seen is the time to market for changes. It's ridiculously fast."

He added that there are financial savings resulting from the increased speed and efficiency of MetLife's data integration process, but he's never been able to put a specific dollar amount on it.

As with MetLife's problems -- lack of speed chief among them -- Palmer said the organization's results after adopting Tibco Data Virtualization are typical.

"Weeks, or even months, to hours is common," he said. "I think of the ROI in terms of agility. There's the reduced cost of doing all the data copying. And one of the other things we hear about is reducing errors that result from moving data around and sharing files, which can lead to mistakes. Reducing that is huge."

Future plans

With MetLife's investment management division meeting success with Tibco Data Virtualization, MetLife is looking at using data virtualization across other divisions too, according to Hammond.

"I did a demonstration for MetLife Bank, and they're looking at bringing on [Tibco] Data Virtualization," he said. "It will be interesting to see what they do with that."