The five Enterprise BI platforms outlined here fall naturally into two camps. The first is the group from traditional large suppliers – IBM, Oracle and SAP. Organizations wedded to one of these suppliers will probably choose their BI offering. The other two specialize in BI and as such, in my opinion, provide superior offerings, although it all depends on what your needs are. Microstrategy has to be congratulated for seeing the inevitable convergence of BI and data mining and supporting data mining models in its BI offering. Spotfire is a fairly unique product quite capable of addressing both BI and data mining needs, and for that reason it has been included.
IBM Cognos is a sprawling set of products aimed at providing a complete BI solution for organizations of all sizes, but specifically large organizations. Unlike some other offerings that incorporate data mining and predictive analytics capability, Cognos is targeted specifically at the traditional reporting and analysis tasks using a variety to presentation techniques. These include dashboards, graphs, charts, reports, tables and so on. For organizations that have not yet implemented an organization wide solution to this need Cognos is a prime candidate. Perhaps the strongest feature is the ability of offer an enterprise wide, integrated BI environment. And you wouldn’t expect anything less from IBM, if does major on infrastructure after all. Having said this Cognos provides a necessary, but hardly sufficient set of tools to gain commercial advantage. It may satisfy the needs of administrative and planning functions, but provides very little for marketing, sales, product development and other functions that are forward looking. For this users will need to embrace IBM SPSS as a separate set of capabilities, or some similar set of tools.
IBM has layered the Cognos products so users can start small and grow. The most basic entry level product is the free version of Cognos Insight. This is a personal reporting and analysis tool, which in the free version will access CSV files (nice as a taster, but not really practical). The paid for versions will access a much broader set of data resources. Cognos Express is targeted at workgroups and midsize companies, and Cognos Enterprise is the full-blown BI and performance management toolset. Over thirty other Cognos products are available to support infrastructure requirements, address specific needs, and provide additional means of data visualization.
What IBM Cognos offers is a platform for looking at what has happened, what is happening, and what might happen in your business. Every organization needs this sort of capability and IBM is one of half a dozen traditional suppliers (Microstrategy, SAP, Oracle, SAS) who offer a solution. However there are new kids on the block in the shape of Qlikview, Tibco and Tableau, although they typically do not have the tailored add-ons of the more traditional suppliers.
In summary there are no compelling reasons to choose Cognos unless the suite of products offers a particularly good fit with needs, or the organization has an existing heavy commitment to IBM products and services.
Verdict: MicroStrategy is one step ahead of much of the competition and delivers a highly configurable and quite uniquely functional BI environment. Should be on any BI shortlist.
MicroStrategy pretty well does everything when it comes to BI and analytics. Data visualization and exploration compares well with best of breed, scorecards and dashboards are feature rich and easy to use, the reporting capability is extensive and, somewhat uniquely for a BI platform, MicroStrategy supports the implementation of data mining models. This does not however mean that MicroStrategy supports the building of such models, because in the main it doesn’t. More on this in a moment.
The MicroStrategy Business Intelligence Platform is described as enterprise class, and indeed it is. It contains too many components to provide details here, but the architecture is extensive and capable of meeting any need..
The range of supported data sources is exhaustive and includes big data. Numerous management, development and reporting tools provide an extremely flexible environment, and delivery is via web browsers, desktop, mobile devices and through other applications such as Microsoft Office.
Scalability is guaranteed trough the MicroStrategy Intelligence Server which supports asymmetric clustering, shared caching and many other features. Security is also sophisticated and robust.
Now back to support for predictive analytics and data mining. MicroStrategy has to be congratulated on its support for PMML (Predictive Model Markup Language), and industry standard for representing predictive models. It means these can be incorporated into reports, visualizations and dashboards so that scoring mechanisms can be seen. The models have to be built using PMML compliant tools, of which there are many. But this is an extra, and much needed dimension to BI and will assume more importance as data mining and predictive analytics become more widely used.
MicroStrategy offers various packages for different needs and size of organization, and it also has some useful free trials. It is very hard to fault MicroStrategy; they have been around a long time, but always seem to keep one step ahead of much of the competition – worth a look.
Oracle BI Foundation Suite encompasses enterprise reporting, dashboards, ad-hoc analysis, multi-dimensional OLAP, scorecards and predictive analytics in one integrated platform. This is a very large suite of products, and Oracle provides ample documentation. So there is little point drilling down into details here, and we will focus on architectural issues.
Visualizations
Data visualization is hot right now and Oracle has responded by adding numerous new visualization modes in its BI toolkit. These include ‘recommended visualization’ for specific data and intent, performance tiles to give prominence to a metric or set of metrics, waterfall diagrams often used in pricing analysis, map views and many, many others. It easily competes with the new generation of visualization tools, but is of course just a single feature of a much broader capability.
Oracle Exalytics
Oracle Exalytics is a combination of hardware and software that facilitates very high levels of analytics performance. This is an in-memory machine with built-in analytics capability that addresses BI, predictive analytics, data mining and Enterprise Performance Management needs. The Oracle BI suite takes advantage of the Exalytics speed, and it is claimed at least that use of Oracle Exalytics can reduce cost of ownership – but we have to take Oracle’s word for that.
Oracle Essbase
Oracle Essbase provides an OLAP solution and can be implemented on the Exalytics platform for performance improvements.
Oracle Exadata
The Oracle Exadata Database Machine is a large scale, scalable data warehousing and OLTP database platform. Exalytics complements Exadata by providing analytics capability.
These various database platforms clearly need to be configured for individual organization requirements. But the diversity of function provides a rich set of configurations, and Oracle is one of only two or three supplier who can match this flexibility.
Other components of the BI suite include BI Publisher. This provides a web based platform for authoring, managing and delivering interactive reports, dashboards and other forms of formatted information.
Oracle BI Suite Standard is aimed at small to medium size organizations, and Scorecards and Strategy Management support execution of strategy by cascading goals down to individual workers and departments.
All-in-all this is a comprehensive BI environment with many, many configuration options. The improved visualization capability keeps Oracle competitive with newer technologies, but within the context of a broader architecture.
Yes, SAP Business Objects (BO) will address your BI needs – no doubt about it. But so will many other products – MicroStrategy, Oracle, IBM Cognos and so on come to mind. Now BO is big. It is big on capability, big on footprint, big on required skill-sets and tends to be big on cost (consultants love it). SAP groupies will undoubtedly want to use BO, others might want to reflect more carefully.
SAP BO comprises over twenty products that address everything from Microsoft Office integration through to data integration. It would be truly futile to provide details on these components, although it is worth pointing out that SAP Predictive Analysis uses the same visual front-end as the rest of the products – and this is a particularly useful feature as the boundaries between BI and data mining start to blur – as they inevitably will.
Criticism of BO usually revolves around complexity and cost. Some of the newer BI entrants (specifically QlikView and Spotfire) provide much more manageable BI environments at a fraction of the cost. But then again they are not quite so all-embracing. As always it’s a trade-off. If you really want a huge monolithic BI environment then be prepared to pay for it. If you don’t, then accept that some of the bells and whistles might be absent – although an extensible BI environment (again QlikView and Spotfire) should make up for this. The choice is yours.
Tibco has got a real winner with Spotfire. It provides an easy to use interface for data visualisation, analytics and the creation of dashboards. Most of the slicing and dicing can be done through drag and drop and a multitude of intelligent functions (eg scaling the time axis on charts automatically) make light work of many analysis tasks. The lightening fast execution speeds are also a great advantage, particularly on large data sets.
More complex analytics can be accomplished through the R programming language, and the R runtime engine has been embedded into the Spotfire statistical server. This allows R based analysis to be fed out to as many users as required (typically through its WebPlayer web client).
Version 5.0 of Spotfire has embraced big data and particularly in-database analysis, with support for Oracle, SQL Server and Teradata – other will follow.
A developer licence costs around US$4000 and the server licence per user is US$500 – with volume discounts.