A New Data Connection From IBM i To Watson Analytics
June 14, 2017 Alex Woodie
Big data analytics is everywhere these days. It simply can’t be helped. And now that IBM is supporting the capability to load DB2 for i data directly into the Watson analytics service, it could be worth your IBM i time to check out what IBM has cooking in Cognitive.
IT Jungle recently checked in with Steve Sibley, IBM’s vice president and business line executive in charge of systems, to talk about the latest developments with Watson, including the new DB2 for IBM i Connect tile on BlueMix.
The new tile “is designed to make it easier for our clients to leverage Watson and connect their IBM i database into Watson,” Sibley says. “That’s a brand new capability . . . we already had clients who were starting to leverage applications, like free form RPG, to connect from an application standpoint into Watson. But now you can actually connect your data securely into Watson as well, to leverage all of the capabilities we have there in that environment.”
BlueMix, if you’re not familiar, is IBM’s API-oriented cloud development platform. The offering is designed to significantly streamline the task of developing cloud-resident applications. Think of it as a largely point-and-click repository for rapidly prototyping new apps that use many of the cloud-based capabilities that IBM and its partners expose on BlueMix.
The fundamental building block in BlueMix is a tile, and users develop their apps by stringing together various tiles in such a manner that the APIs are linked behind the scenes. And now one of these tiles is designed to pull your DB2 for i data off your on-premise Power Systems box and upload it, via HTTPS, into the BlueMix cloud so your analysts can “play” with it using Watson analytics tools and run analytic applications against it.
“We’re leveraging the Watson services to do analytics against your secure transitional data you may have in your DB2 for IBM i environment,” Sibley says. “What the connector does is it provides the ability for the Watson service in the cloud to securely go through an HTTP secure server directly into DB2.”
The new DB2 for IBM i Connect tile does the job of an extract, transform, and load (ETL) tool for you, except that it can also run in a real-time manner, Sibley says. “It will enable you to . . . leverage your own secure, on-premises core business data that you have in DB for i in a Watson services application that you develop,” he says.
While the new IBM i Connect tile brings in structured data from the DB2 for i database, other BlueMix tiles can help a customer tap into unstructured data sources, such as social media posts or Web comments. Analyzing this mix of data will give IBM i shops a better perspective on where they stand now – and in the future – compared to just analyzing the structured data in the relational database.
“So it’s being able to say, ‘This is what I’m hearing is going on in LinkedIn or Twitter or whatever it may be,'” Sibley says. “How does that match up to what I’m seeing in transactional data? How does that match up to what I have as far as inventory goes? And do I need to do something to respond to what I hear is going on in the marketplace?”
This big-data style of analytics is not easy, and programming it is a decidedly different task from the procedural logic that dominates in RPG shops. While data science is the popular term for it today, some people still like to call this sort of thing “applied statistics,” because it’s trying to spot correlations and anomalies, and do something interesting with them.
Many organizations today are looking to hire data scientists who have the mathematical chops to handle advanced statistics’ the technological chops to express those statistics in Python, R, or other languages; and the business chops to know what metrics matter.
Of course, companies like IBM are chipping away at those big data analytics requirements, and are aggressively working to automate many of those data science engineering tasks using the power of software. That’s not to say that IBM has hidden all of the complexity yet, but it’s definitely lowered the barrier to entry, and that’s an important thing to realize for organizations that have stood on the big data sidelines up to this point.
“It does take some expertise to program, depending on the level of detail and the program you provide,” Sibley allows. “But we do have some people who have gone and done some stuff fairly quickly. These are not PhD types of programs from an application or customer standpoint. They can go and develop applications using Watson in a fairly short timeframe.”