April 11, 2017
The recent announcement of IBM Machine Learning on z/OS
is significant for a couple of reasons. First, it's another example of IBM's ongoing investment to the mainframe. Beyond that, it's important for anyone who supports software installation, database administration and analytics on z/OS.
Studies have shown that 70 percent of the world's data being accessed for analytics originates on the mainframe.
While one of the traditional reasons for moving the data off z/OS has been to avoid impacting operational online transaction processing (OLTP) performance with resource intensive online analytic processing (OLAP), over the past few years DB2 for z/OS has evolved into a hybrid transactional/analytical processing (HTAP) database.
DB2 for z/OS--combined with IBM DB2 Analytics Accelerator--form a self-managing database management system that maximizes the efficiency of each workload it runs. Because of this, each transaction--OLTP, batch, analytic query--is executed in its optimal environment, providing the best performance and greatest cost efficiency. Supporting transactional and analytic processing in a single environment allows organizations to infuse DB2 for z/OS data into business-critical analytic applications while providing fast, accurate responses to analytic queries and predictive scores. This gives data analysts a holistic view of enterprise data never before possible.
To support this direction, IBM is bringing critical open source applications such as machine learning and Spark to z/OS. Machine Learning for z/OS is being used to infuse continuous intelligence into the enterprise to improve the productivity of data science teams.
As for Spark, you can read about it in this two-part article called "Machine Learning Using DB2 for z/OS Data and Spark." Part 1
shows how the Vector Assembler is used to create features as input. Part 2
shows how to use R formula to solve the problems introduced in part 1. DB2 for z/OS Modern Applications
, an IBM developerWorks community, is another great resource.
If you're reading this, you certainly understand that the long-term viability of the mainframe has never looked better. Again though, this is another example of the mainframe not only adapting to but facilitating the latest technologies. It's exciting stuff.
Posted April 11, 2017 | Permalink