IBM's zAnalytics strategy provides a complete picture of analytics on the mainframe using DB2, the DB2 Analytics Accelerator, and Watson Machine Learning for System z. The presentation discusses updates to DB2 for z/OS including agile partition technology, in-memory processing, and RESTful APIs. It also reviews how the DB2 Analytics Accelerator can integrate with Machine Learning for z/OS to enable scoring of machine learning models directly on the mainframe for both small and large datasets.
20. 20
IBM Machine Learning for z/OS
Integrating the DB2 Analytics Accelerator with ML for z/OS
IBM Machine Learning for z/OS
ML for z/OS enhances the analytics
solution as follows:
Providing tooling for entire data
scientist life cycle:
• Model build, train, validation, ...
• Health monitoring
• Model optimization
Enables scoring on z/OS platform
Leverages Spark MLlib and Spark for
z/OS as runtime engine
. . .
IBM DB2 Analytics Accelerator
DB2 Analytics Accelerator enhances
the analytics solution as follows:
Supporting data engineering tasks:
• z/OS data can be prepared via in-
DB transformation (AOT usage)
SQL access to z/OS data can be
accelerated
HPSS archiving can be used to
accomodate huge amount of z/OS
data
. . .
21. 22
IBM Machine Learning for z/OS
DB2 IDAA
data data
asynchronous
replication
most recent
committed
data
available?
yes
no
Write
requests
OLTP read
requests
OLAP read
requests
wait for
given
time
period
most recent
committed
data
required?
yes
no
initiate
apply
Reading most recent committed Data during asynchronous Replication
called ZERO LATENCY
22. 24
IBM Machine Learning for z/OS
DB2 Analytics Accelerator & ML for z/OS – Complementing Values
Situation #1
Small amount of z/OS
data
zIPP / memory with
sufficient capacity
Scala, Python for data
prep required
Jupyter Notebook of
ML for z/OS used
No SQL skills or SQL
not appropriate
...
ML for z/OS
DB2 Analytics AcceleratorDB2 Analytics AcceleratorDB2 Analytics Accelerator
ML for z/OS ML for z/OS
Situation #2
Large amount of z/OS
data
DataStage (or similar
tool) already used
SQL accepeted for data
prep
Z capacity insufficient
for data prep
ML for z/OS Jupyter
notebook used for
addtl. data prep (similar
to SAS data cube build)
Supporting broader
data lake topologies
Accomodating more
data due to archiving
Need for limited R
support
...
Situation #3
Large amount of z/OS
data
DB2 Analytics
Accelerator already
deployed
Addtl. points similar to
situation #2
Need for limited R
support
...
Situation #3
Large amount of z/OS
data
PMML needed
Batch scoring
SPSS Modeler and
SPSS C&DS already
used (integrates with
the Accelerator)
Interest in in-DB
Analytics
Need for limited R
support
Supporting broader
data lake topologies
...