Posts Tagged ‘predictive analytics’

Real Time Analytics on the IBM z13

June 4, 2015

For years organizations have been putting their analytics on distributed platforms thinking that was the only way to get fast, real-time and predictive analytics. Maybe once but not anymore. Turns out the IBM z System, especially the z13 not only is ideal for real time, predictive analytics but preferable.

IBM today is so bullish on analytics, especially predictive analytics, that last month it introduced 20 pre-built industry-specific predictive analytics solutions. To build these solutions IBM tapped its own experience working on 50,000 engagements but also an array of outside organizations with success in predictive analytics, including Urban Outfitters, National Grid, Deloitte, Bolsa de Santiago, Interactive Data Managed Solutions, and Bendigo and Adelaide Bank, among others.

Web

Courtesy of IBM (click to enlarge)

The truth of the matter is that without efficient real time, predictive analytics managers get it wrong most of the time when it comes to making operational decisions, said Paul DiMarzio, IBM z Systems Big Data and Analytics Worldwide Portfolio Marketing Manager. He spoke at IBM Edge2015 in a session titled When Milliseconds Matter: Architecting Real-Time Analytics into Operational Systems. His key point: you can do this completely within the IBM z System.

The old notion of sending data to distributed systems someplace else for analytics now appears ridiculous, especially with the introduction of systems like the z13 that can handle operations and perform real time analytics concurrently. It performs analytics fast enough that you can make decisions when the action is still going on. Now the only question is whether we have the right business rules and scoring models. The data already are there and the tools are ready and waiting on the z13.

You start with the IBM SPSS Modeler with Scoring Adapter for zEnterprise. The real time predictive analytics capability delivers better, more profitable decisions at the point of customer impact. For business rules just turn to the IBM Operational Decision Manager for z/OS, which codifies business policies, practices, and regulations.

IBM SPSS improves accuracy by scoring directly within the transactional application against the latest committed data. As such it delivers the performance needed to meet operations SLAs and avoid data governance and security issues, effectively saving network bandwidth, data copying latency, and disk storage.

In addition to SPSS and the Operational Decision Manager the z13 brings many capabilities, some new for the z13 at this point. For starters, the z13 excels as a custodian of the data model, providing an accurate, secure, single copy of information that, according to IBM, ensures veracity of the data necessary for reliable analytics and provides centralized control over decision information.

Specifically, the machine brings SIMD (single instruction multiple data) and the MASS (mathematical acceleration subsystem) and ATLAS (automatically tuned linear algebra software) libraries for z/OS and Linux on z. SIMD enables the same operation to be performed on several data elements at the same time rather than sequentially. MASS and ATLAS help programmers create better and more complex analytic models.

In addition, increases in memory to as much as 10 TB, faster I/O, and simultaneous multi-threading (SMT) generally boost overall throughput of the z13, which will surely benefit any analytics being run on the machine, especially real time, predictive analytics.  In addition, analytics on the z13 gains from deep integration with core systems, the integrated architecture, and its single pane management view.

The latest IBM Red Book on analytics on the z13 sums it up as such: z Systems analytics enables organizations to improve performance and lower cost by bringing the analytic processing to where the data resides. Organizations can therefore maximize their current IT investments while adding functionality and improved price and performance with the z13. And with the new z13 features, applications can gain increased throughput for operational business intelligence (operational BI) and DB2 query workloads, which saves money (hardware, software, labor).

The Red Book suggests the following example: a user with a mobile application signs on and initiates a transaction flow through an IBM MobileFirst Platform Server running on Linux on z. The event goes to an LDAP server on z/OS to validate the user’s sign-on credentials. After successful validation, the transaction then proceeds through the z/OS transaction environment where all of the data resides in DB2 z/OS. IBM CICS transactions also are processed in the same z environment and all of the analysis is performed without moving any data, resulting in extremely fast performance. Sweet.

DancingDinosaur is Alan Radding, a veteran IT analyst and writer. Please follow DancingDinosaur on Twitter, @mainframeblog. See more of his IT writing on Technologywriter.com and here.

Predictive Analysis on zEnterprise

January 9, 2012

IBM has been positioning the System z for a key role in data analysis.  To that end, it acquired SPSS and Cognos and made sure they ran on the z. More recently, growing interest in Big Data and real-time data analytics only affirm IBM’s belief that as far as data analytics goes the zEnterprise is poised to take the spotlight. This is not completely new; DancingDinosaur addressed it in October 2009.

Over the last several decades people would laugh if you suggested a mainframe for data analysis beyond the standard canned system reporting.  For ad-hoc querying, multi-dimensional analysis, and data visualization you needed distributed systems running a variety of specialized GUI tools. In addition, you’d want a small army of business analysts, PhDs, and various quants to handle the heavy lifting. The resulting queries could take days to run.

In a recent analyst briefing, Alan Meyer, senior manager for Data Warehousing on z, built the case for a different style of data analysis on the zEnterprise. He drew a picture of companies needing to make better informed decisions at the point of engagement while applications and business users are demanding the latest data faster than ever. At the same time there is no letup in pressure to lower cost, reduce complexity, and improve efficiency.

So what’s stopping companies from doing near real-time analytics and the big data thing? The culprits, according to Meyer, are duplicate data infrastructures, the complexity of integrating multiple IT environments, insufficient and inconsistent security, and insufficient processing power, especially when having to handle large volumes of data fast. The old approach clearly is too slow and costly.

The zEnterprise, it turns out, is the ideal vehicle for today’s demanding analytics.  It is architected for on-demand processing through pre-installed capacity paid for only when activated and while adding processors, disk, and memory without taking the system offline.  Virtualized top to bottom, the zEnterprise delivers the desired isolation while prioritization controls lets you define critical queries and workloads. Its industry-leading processors ensure the most complex queries run fast, and low latency enables near real-time analysis. Finally, multiple deployment options means you can start with a low-end z114 and grow through a fully configured z196 combined with a zBX loaded with blades, especially the IBM DB2 Analytics Accelerator (IDAA), a revamped version on the Smart Analytics Optimizer.

Last October IBM unveiled the IDAA and a host of other analytics tools under the smarter computing banner. But the IDAA is the zEnterprise’s analytic jewel. There IDAA incorporates Netezza, which speeds complex analytics through in-memory processing and a highly intelligent query optimizer. When run in conjunction with DB2 on the z, the results can be astonishing, with queries that normally require a few hours completed in just a few seconds, 1000 times faster according to some early users.

Netezza, when deployed as an appliance, streamlines database performance through hardware acceleration and optimization for deep analytics, multifaceted reporting, and complex queries. When embedded in the zEnterprise, it delivers the same kind of performance for mixed workloads—operational transaction systems, data warehouse, operational data stores, and consolidated data marts—but with the z’s extremely high availability, security, and recoverability. As a natural extension of the zEnterprise, where the data already resides in DB2 and OLTP systems, the z is able to deliver pervasive analytics across the organization while further speeding performance and ease of deployment and administration.

Already companies are reporting valuable results. Marriott Hotels reports using the system to book inventory down to the last room available to maximize yield. Chartis Insurance turned to it to meet SLAs that allow for no down time while requiring high performance and fast time to market. In the process, it achieved what it reports as seamless 99.99% up time, the fastest performance available, and time to market measured in days. Swiss Re turned to IDAA to put the right answers into the hands of decision makers across the business.

Today, IDAA and Netezza are just two components of a comprehensive zEnterprise data analytics portfolio that includes Cognos, SPSS, InfoSphere, Guardium, Optim, QMF, MDM, and more. IBM offers powerful data analytics capabilities for its Power and System x platforms, but the IDAA, which is the heart of the fast, near real time predictive analytics, is available only for the zEnterprise, either the z114 or the z196. Might be ideal for a private analytics cloud.


%d bloggers like this: