In the business intelligence world at large, predictive analytics and "big data" are being touted as cutting-edge concepts. Although the IBM i was behind in these areas for awhile, it's quickly catching up.
In May 2011, IBM published a paper entitled "The Essential CIO – Insights From the Global CIO Study," an overview of the results of interviews with more than 3,000 corporate information officers around the world. In a nutshell, that study showed that the top concern (83 percent) of those surveyed was "business intelligence and analytics." While one might object that that topic would only naturally be the concern of the typical CIO, it's also true that in the nearly two years since that document was published, business data and the information to be retrieved from it has only grown in importance.
For many of the SMBs that make up the largest slice of the IBM i market, business intelligence (BI) has traditionally been the process of analyzing what has already happened: how much of what products and services have been sold to what customers, how well this year's financials are stacking up against those of previous years, and similar concerns. Query tools that answer specific questions from business databases, dashboards and other graphical tools that track key performance indicators (KPIs), and report-writers that put the good (or bad) news into succinct packages for quick digestion by decision-makers are the trusty hallmarks of this kind of BI.
However, for larger enterprises and more forward-looking SMBs, simply knowing where you've been isn't good enough. Equally important is the ability to wisely plan where you're going. That's where predictive analytics (PA) and "big data" enter the picture.
PA and Big Daddy Data
PA is the process of using information to forecast the likelihood of future events or actions—for example, how probable it might be that a past customer will make future purchases and what those specific purchases might be. Armed with this knowledge, the enterprise could then take targeted actions to increase the likelihood of such a repeat purchase, or facilitate it when it takes place. PA uses models that, generally, assign a value called a predictor to the hoped-for future event and then use behavioral data and other patterns extracted from available information to guide business decisions going forward. Examples of using this kind of data would be restaurants deciding which less-popular items to replace on their menus and franchise operations deciding which underperforming stores to close based on projected sales volume.
Big data is a concept that involves using very large data sets for BI, PA, and other kinds of data analytics in business, science, and other fields. IBM's own Big Data pagcites examples such as analyzing streaming data about medical patients to decrease patient mortality and decreasing telephone-call processing time by examination of networking and call data. The trouble with big data, of course, is an enterprise's ability to process data. Analyzing up to about a terabyte of data is doable for most enterprises, but when potential data sets pass hundreds of terabytes enroute to multiple petabytes, the number of servers needed to effectively process that much data can become prohibitive.
When it comes to data, like many things, our culture tells us that more is usually better. Logically, it follows that the more data you can analyze to get an answer, the more likely that answer is to be a better one than you'd get by looking at less data. The problem is where to draw the line between enough and too much, and when it comes to advanced data functions like PA, perhaps that line is currently beyond the scope of many IBM i user companies.
When Too Much Still Isn't Enough
"To be brutally honest," notes Alan Jordan, senior vice-president of operations at Coglin Mill, makers of the RODIN data-warehousing product for IBM i and other platforms, "in my opinion very few IBM i shops are getting to the level of sophistication needed for PA. PA tools such as IBM's SPSS Modeler rely on carefully prepared data. This can be a very difficult and time-consuming task. Each 'model' used in PA will likely need the data to be prepared in a different way. This takes a lot of time and effort if you don’t have the right tools."
On the other hand, maybe PA isn't as important to most SMBs as it might be to larger organizations.
"Most [of our customers] are what IBM calls small to mid-size companies with somewhere between 50 and 500 employees," reports Bill Langston, marketing director for NewGeneration Software, which produces multiple business-intelligence and dashboard products for the i. "In that customer segment, we are rarely asked about predictive analytics, so at least for our typical customer, I'd have to say PA doesn't drive much BI demand on IBM i. Most customers we speak to are very focused on improving operational reporting and increasingly looking for ways to extend data access to mobile users' smartphones and tablets."
"The first step on the journey into PA is to design and build a data warehouse," counters
"And we can't talk about critical technologies in BI without mentioning big data," he adds. "To a certain segment of IBM i shops, it is already relevant and important, but if you haven't implemented traditional BI over structured data with a data warehouse, then you can't take the leap into big data."
Of course, whether or not an enterprise considers moving down the road to data warehousing, there are still significant issues with data quality and data accessibility, obstacles that
"Reporting and analytic tools will not identify the errors in your data. Believe me, every database contains bad data. If you don't know how reliable it is, how can you trust the reports produced using it? Bad data is a significant issue for virtually every organization and is estimated to cost
"Another critical issue is that these days we don't just have data in our IBM i application databases," he explains. "We have it in MS SQL Server, Oracle, and other databases. We have it in cloud applications such as Salesforce. Bringing all of this data together for BI reporting can be a huge challenge. It may be related, but it is formatted differently. Joins are impossible. Duplicates exist, and there are all sorts of other challenges. A data warehouse solves this issue."
Looking to the Cloud for Help
Laying down guidelines for determining the point at which data warehousing becomes feasible for enterprises of varying sizes is beyond the scope of this article, but perhaps cloud technology is a long-term answer to the challenge of putting such tools as PA within the reach of enterprises using the i.
"A private cloud approach lets the customer scale their use and investment by user, by month, for a specific allocation of disk space, memory, and CPW," advocates Langston. "We think it will be come a very popular way to extend data access to more IBM i customers in the next few years. We also expect to see a rise in the use of very low-cost cloud-based analytical tools that will enable customers to achieve results more cost-effectively and with far less implementation consulting. I think companies who are about to commit to a large BI project with a major vendor should think again about their plans to make sure they aren't making an unnecessary investment to obtain capabilities that will soon be matched by some of these new and inexpensive cloud-based solutions."
Being Clever with What You Have
Regardless of whether or not PA, big data, or data warehousing is in your enterprise's future, making the most of the resources you already have is crucial.
"One of [our] most innovative customers is providing critical information from their ERP systems here in the
"One of our long-time customers uses their data warehouse as a clearing house of data for many purposes, way beyond the range of traditional BI. On a daily basis, [the warehouse] prepares general ledger interface data from their IBM i applications and pushes it to Oracle. Similarly, payment information is sent to their accounts payable system to print checks. It also drives their Web applications, and the list goes on," reports
Available BI Tools for IBM i
What follows is a list of the major decision-support products currently available for IBM i. The first section shows products that aren't fully fledged BI systems but provide significant data-analysis and display capabilities for budget-minded enterprises that still need to get a better handle on their business data. The second section shows more comprehensive BI and data-warehousing offerings. Not included are report-design tools, products that simply handle reporting, and applications primarily designed for accounting, financial, and ERP functions, even though those apps fulfill many BI functions.
Database query tools are also not included here, despite their relevance to BI. Please see "Technology Focus: Basic Database Query Tools Still Have a Place at the Table" for an overview of those kinds of software applications for the IBM i.
The listings provide only a brief overview of each product's functions but include links to product Web pages where more complete information is available.
And as always when looking for products or services, be sure to check the MC Press Online Buyer's Guide.
Dashboard and Intermediate Products for IBM i
Attunity CDC captures and highlights all changes made to any corporate data sources, including databases, data warehouses, and non-relational data sources. Data sources can reside on IBM i and z systems, HP Nonstop, OpenVMS, UNIX, and Windows platforms.
Connect provides an array of database adapters that connect to a variety of PC, midrange, and mainframe databases and let users make this data available to any Java, .NET, and Web applications. Connect offers universal SQL database connectivity and XML data services for relational sources such as DB2/400, DB2, and Oracle, as well as non-relational sources such as text files.
Federate lets users build single views of corporate data from the same range of sources as its CDC and Connect products.
Tracker Data Warehouse is a browser-based data miner that uses SQL Server as a data store and extracts information for display in spreadsheets, one of 40 preconfigured report templates, or user-designed custom report formats.
SEQUEL is a query and reporting tool that delivers data to the user's choice of graphical interfaces, such as dashboards, tables, custom-formatted reports, Web displays, and other media. It also provides options for data cleansing and reformatting, data drill-down, and distribution via email and other means.
IBM's Cognos Real-time Monitoring displays KPI and other business operational data on self-service dashboards to augment other BI data.
SPSS Collaboration and Deployment Services is a consultation service that helps enterprises share data assets across multiple departments, enterprises, or other entities. Services include automating analytical processes and deployment of analytic results in existing or custom-built applications and other venues.
LANSA Client helps non-technical users query, report, and analyze data from IBM i and Windows servers. Users can download the data to MS Excel spreadsheets, graphical views such as charts and graphs, and comma-delimited ASCII text files.
Logi Info lets non-technical users build custom reports, dashboards, Web portals, interactive mobile apps, and other analytical tools using nearly any data source. The product includes more than 600 pre-built elements that can be selected to construct data display and analytical tools without programmer help.
Momentum's inTuator offers multidimensional analysis and charting tools for existing IBM i data. Users can link data across application areas, work with downloaded data offline, and export or download data to MS Office applications.
Business Performance Dashboard lets business professionals view, present, and interact with charts, reports, tables, gauges, calendars, maps, and other visualization aids via a Web browser. Information from these sources is also downloadable to MS Excel spreadsheets without additional programming.
SmartView is a Windows-based, online analytical processing (OLAP) application that uses IBM i data gathered by the NGS-IQ application (see below). SmartView lets users analyze and display data in a wide range of ways, as well as customize it to show specific information types via drag-and-drop formatting aids.
Web QL is a proprietary language that runs on servers using UNIX or Windows and can harvest nearly any kind of data from Web and non-Web sources (including SQL), handle data transformation and restructuring processes, and then download the collected data to existing internal and external systems.
NetCharts lets executives and other nontechnical users select operational business data and generate dashboards that display and analyze KPIs, reports, and other business metrics.
Full-Service Business Intelligence Products for IBM i
Andrews Consulting Group
RapidDecision is a BI system that includes pre-built data warehousing and datamart solutions that work with existing enterprise resource planning (ERP) applications to provide data analysis support.
Business Computer Design International (BCD)
Nexus Portal provides a prefabricated, customizable Web portal for accessing enterprise information on IBM i machines. Nexus Portal can display information on Internet or intranet portals, provides security features, provides dynamic and menu navigation, and can generate dashboards that focus on KPIs.
RODIN Data Asset Management is a data-management environment for data warehouses, datamarts, and other data-integration applications. It includes tools for describing and managing metadata, acquiring new data, handling change-management and versioning, and managing ETL processes. The product is available in four editions tailored to different-sized organizations and includes an optional DB2 Web Query version.
With its 2011 acquisition of Oco, Inc., Deloitte offers BI analytic applications that can draw data from application systems such as Lawson and SAP, as well as homegrown System i-based applications. From this data, DMA Managed Analytic Applications can help users build dashboards, scorecards, user-specific reports, and other aids for analyzing business data.
An end-to-end BI solution, Diver includes analysis, integration, and dashboard features, as well as security, real-time alerts, and extensive deployment options. It lets users build and deploy customized BI options using data from IBM i and other data sources.
Cognos BI provides reports, analysis, dashboards, and scoreboards to help users capture business data and other metrics to provide a well-rounded view of enterprise activity. It helps users analyze information as well as distribute it via impromptu or standardized reports, dashboards, and other graphical means, including social networking.
ShowCase Warehouse Builder generates a data warehouse or datamarts from corporate data on IBM i servers, as well as helps handle data transformation, loading, replication, and management.
The iWay Integrity Solutions suite provides several products that underlie a strong business intelligence system. These include data quality management (for optimizing data correctness), master data management (for consolidating up to millions of records), and data governance (management and control of data by its internal owners).
Performance Management Framework can draw information from more than 300 IBM i and other platform data sources to deliver enterprise performance management information to all levels of business executives. The product includes more than 500 pre-packaged metrics for common business functions and can produce information in a wide array of graphical formats.
WebFOCUS Business Intelligence provides business data for analysis via dashboards, scorecards, query and analysis tools, mobile devices, guided ad hoc reporting, integration with desktop products, and dynamic report distribution with real-time alerts.
WebFOCUS RStat is a predictive analytics and data-mining application for corporate data. It offers a single platform for BI, data modeling, and data scoring that helps users build and export models for prediction and classification, and provides data exploration, descriptive statistics, and interactive graphs.
Visual Discovery is a data-analysis solution with built-in visualization features. It lets users build multidimensional, graphical representations of data and output completed dashboards to MS PowerPoint, Adobe PDF files, and compound WebFOCUS reports.
Decision Assist is made up of prefabricated business analytics and reporting models that specialize in either financial performance or business performance. Designed specifically for i5/OS, it includes a built-in datamart, a Web portal user interface, dashboards and multidimensional models, and predefined spreadsheets and reports.
NGS-IQ is a suite of business intelligence products that includes Decision Assist, IQ Client (a Windows query and Web report developer), Qport Access (data transfer to DB2 databases and MS Office apps), Qport Builder (a Web-based query developer), Business Performance Dashboard (browser-based graphical displays for business data), and IQ Server (a 5250 interface for building queries and interfacing with mobile devices).
NGS-IQ Cloud provides cloud-based BI services built on the company's NGS-IQ product suite. Features include security, query, analytics, reporting, and data replication.
SAP's products include a portfolio of BI applications that work with DB2 on all IBM platforms, as well as Lotus Notes/Domino, WebSphere portals,
Tailored for manufacturing operations, Stratum offers the ability to analyze data gathered in enterprise resource planning (ERP), customer relationship management (CRM), and Web-based transaction applications. It lets users compare and rank KPIs from different business areas, access hundreds of pre-built views and reports, carry out predictive and budget activities, and deliver information throughout the enterprise.