[This post originally appeared in Oracle Magazine]
By Jeff Erickson
October 4, 2019
Charlie Berger wants DBAs and database developers to click a link in their autonomous database service console that can open a new world in their careers. It’s there alongside the common data management options for downloading clients or setting passwords, and it’s called Oracle ML Users. “You click on it and now you’re going down a different hallway to the land where the data scientist people hang out,” says Berger, a senior director of product management for machine learning, AI, and cognitive analytics at Oracle. Berger, a 20-year veteran of data mining and machine learning in Oracle Database leads sessions and hands-on labs at Oracle and user group events around the country.
With Oracle Autonomous Database, when DBAs click the Oracle ML Users link, they’ll find Oracle Machine Learning-based notebooks where they can define business problems, gather and prepare data, and apply machine learning algorithms available from an extensive library. Next, the DBA, database developer, or analyst can use a simple Zeppelin notebook user interface to build and test machine learning models. Much of the work will be familiar to database experts. “Database developers perform many of these tasks already,” Berger says, “but they refer to them as ETL [extract, transform, and load], data wrangling, and moving scripts into production.”
For DBAs, this is an interesting and profitable area to explore with time that’s been freed up by using the autonomous database from Oracle. Autonomous databases deploy, tune, patch, and secure themselves with no human intervention, so a DBA can think about doing bigger and better things. DBAs know their data and know their business, and “with a little coaching in machine learning concepts, they can start building and applying predictive models to their data to help their organizations run more intelligently,” Berger says.
Berger’s experience tells him that the playing field is tilted in favor of DBAs who want to move in this direction. “In my travels, I’ve discovered that it’s easier to take the person who knows and likes SQL and teach them how to begin doing real machine learning in the database, than it is to take folks who do Python or R, and who know algorithms, and teach them about SQL to do machine learning inside the database.”
Berger believes more and more data science will happen inside the Oracle database for the simple reason that “Algorithms are small and data is big. Oracle chose to move the algorithms to the data and not the other way around,” he says. “Because we moved the algorithms inside the database, you don’t have to move ever growing amounts of data out to somebody else’s open source code and algorithms.” This simplifies the process and ensures that machine learning models move immediately into production.
That’s why Berger works to help DBAs get familiar with standard practices in data science: “You have to know what the business problem is, you have to think about your data, you have to transform the data, you have to build the models, and then you have to evaluate the models and finally deploy them.” All of that is available with Oracle Machine Learning in the database.
Berger is excited to show DBAs and database developers the power of the Oracle Machine Learning tab in Oracle Autonomous Database—giving them a ready path to bring their data expertise and their database knowledge to the highly valued business function of data science. “It’s time for DBAs to get a little more of the bright data science spotlight,” Berger concludes.
LEARN more about machine learning.
LEARN more about Oracle Machine Learning for SQL.