(US and Canada) Ricardo Crepaldi, Director of Business Intelligence, Big Data and Integration, BASF, Germany, talks to Dr. Michael Stonebraker, Founder, CTO, and Visionary, Hopara. 

Stonebraker shares that he knows a big pharma company that hired a CDO a few years ago, and he refused to take the job unless he got read access to everything. The board of directors grumbled and then said OK, so then he had enough power to develop a global data integration strategy. He did the same by generating an architecture because data integration involves ingesting, cleaning, and removing duplicates. It’s a whole bunch of moving parts. So he came up with an architecture that could be used throughout the enterprise. 

Now they are moving to basically everybody using that architecture. This means they’re all using a standard set of tools so that he has enough control that he can do the highest value data integration tasks first, and the mantra is data integration. You’re almost certain to fail if you try to ‘boil the ocean’ as your first project, he continues. So, the key is to start small with something that has obvious ROI so that the executives don’t cancel while you’re delivering. Hence, Stonebraker’s mantra is to start small, be successful, then grow with a common architecture over a fair hunk of the enterprise of this big pharma.

He notes people should keep in mind that many enterprises have in-house legacy tools doing a bad job with various data subsets. The people running these ops are engineers. These legacy systems are partial to their continuing to exist. He thinks one has to view this as a business problem and not as a complete employment act for IT workers in the enterprise. Hence, you use the best tool you can get a hold of. He notes that he can guarantee these are not the ones built in-house. 

Stonebraker shares that if one can move to what’s called data ops, basically that says, take your giant monolith systems and divide them up into bite-sized pieces that can be run one by one.

If, to the extent that one can make changes to a legacy system, you should re-architect it as a bunch of micro-ops or just a bunch of smaller ones that may make the pieces bite-sized. it is a good idea for everybody, he advises, and you should be pushing this approach hard if you’re the CTO of an enterprise.

“We need to keep the lights on. Chances are your best people are allocated to doing that sort of grunt work. So, you’ve got the best people doing uninteresting stuff," Stonebraker says. “That’s not what you want to do with your best people. You want them to figure out where the enterprise  should be going, not dragging along this big cinder block of the sins of your predecessor. So you’ve got to empower your best people to do forward-thinking. If nothing else, at least allocate 5% of your IT budget to doing new stuff. If you want to do this in a skunkworks, free up one day a week for your employees to write proposals, work on something new and different.”

If you outsource anything, Stonebraker advises, outsource keeping the lights on and put your best people on the shiny new stuff that you would otherwise outsource. He stresses it’s important to empower your internal people because they will stay and have a much more rewarding job.