Why the Mainframe Rules for Big Data Analytics

ETL -- extract, transform and load -- has been the prevailing mantra for the past 30 years when it comes to data analytics. Until recently, it was just too expensive to run analytics where the data lived, and so companies were forced to move the information to another platform.

But these days the tides have turned: The quantity of data is so vast that it's becoming impractical to transfer the data off the mainframe for analysis.

Agility is absolutely critical to a successful business strategy these days. A key component of agility is the ability to analyze data on an as-needed basis, said Greg Willhoit, of Rocket Software, during his session on analytics at SHARE in Pittsburgh.

"Our bosses need to make quick decisions," he said. "The more agile they are, the greater advantage they have over the competition."

Indeed, business intelligence and analytics are top IT priorities, according to Gartner research that points to more than $1 billion in software sales connected to those areas in 2013.

Because of z System’s capacity, companies in the financial, communication and transportation sectors, among others, still rely on it for their storage needs. ETL worked fine when they had manageable amounts of data to move and analyze off the mainframe.

Now companies receive data in such great quantities from such varied sources -- social networking, web transactions and their own processes, just to name a few -- that Willhoit said ETL can no longer keep up.

"I don't think we can continue to survive using ETL," he said. "It takes too long. It's not very cost effective. We work hard to keep data secure on z Systems. And then what happens when you ETL it to 500 different places?"

The solution, he said, is to virtualize the data and leave it on the mainframe. It's much cheaper to run workloads on the mainframe than it used to be. And companies are already finding success virtualizing data in place.

"We need to rethink the way we've been doing business if we want to be agile," Willhoit said. "Chalk me down as ETL sucks."

Want more insights into the latest mainframe trends? Come to SHARE in Seattle, March 1-6, for six days of robust educational sessions, networking and opportunities to influence the industry. 

Recent Stories
Message From SHARE: Influence in 2021

Microlearning with SHARE Nibbles

Use Zowe to Broaden Assembler Development Practices