FREE email newsletters 
Key Considerations for Adding Business Intelligence to Your IT Infrastructure
InfoManagement Direct, September 2007
Business intelligence (BI) has matured and is almost universally accepted as a critical part of overall IT strategy. It continues to grow in sophistication and technical requirements, and is being used more and more across the enterprise. The real-time analysis it provides enables companies to improve ongoing business processes, maximize revenue and reduce costs. It helps IT departments mitigate risk and develop compliance processes for regulatory mandates such as Sarbanes-Oxley. Most importantly, BI allows for the in-depth analysis of customer data so that management teams can better monitor corporate performance and sales activity. According to Brian Dooley of the Cutter Consortium, "The need to retain information leads to a desire to use it, and data mining moves quickly to data analysis and BI in the quest to leverage information stores."1
Advertisement
A BI Solution for Every IT Infrastructure
No two infrastructures are the same. Some are organized into silos of information that don't talk to each other at all. Others are centralized, where all the data is fed into one database. Some use an enterprise information integration (EII) system to integrate separate data sources together. Still others rely on a service-oriented architecture (SOA). And that doesn't include the data stored in a hosted software as a service (SaaS) application.
But that's only the tip of the iceberg. A single organization can have a mixture of all of these. For example, there may be a set of servers that feed data into a single database alongside a few systems such an on-demand customer relationship management (CRM), which can run independently. A company might also acquire another company that maintains a completely separate data management system. For a broader view of what's going on across the enterprise, IT personnel must integrate data collected across these multiple infrastructures, which requires tedious data entry, labor-intensive extract, transform and load (ETL) tools and the manual combining of disparate reports. Clearly, this is an inefficient method that wastes time, money and resources.
New BI solutions are aimed at analyzing data across the enterprise by working with and linking all types of environments - the siloed, the centralized and the hosted. Ideally, authorized users should be able to access data from any system across the various infrastructures and obtain an aggregate view of that data for comprehensive analysis. The ability to pull data from all sources across the enterprise offers big rewards, including faster, more thorough analytics, better business agility and reduced costs.
When evaluating BI solutions, consider the following:
- How does the system talk to data across the enterprise?
- What type of security infrastructure is supported?
- How is data presented?
- How does the system integrate with your existing infrastructure, and what are the benefits of purchasing an open source versus a proprietary solution?
Accessing, Integrating and Aggregating Data
The volume of enterprise data is ever increasing. The more capable we are of collecting data, the more we have, and that creates the bigger challenge of organizing and using that data. Especially when systems are operating independent of each other, making sense of collected data can be a logistical nightmare. Often, companies will dedicate staff to data entry. They must gather data from disparate systems and consolidate it manually into Excel spreadsheets or other analysis tools. If resources are not available, the data remains separate. Reports from each of the siloed systems often cannot be combined, and the data cannot be compared and manipulated in an efficient way.
BI addresses this problem by providing the capability to combine data from multiple sources, either as part of a batch process or on demand, so that it can be subject to more accurate, in-depth analysis. For example, sales data can be combined with inventory to examine ratios or establish customer behavior patterns.
Given the variety of data sources in today's complex enterprise IT infrastructures, it's critical that the BI solution implemented be able to talk to various data sources and aggregate data from these sources into a single console or reporting server. Manually creating a data warehouse or BI universe has become an antiquated method as new solutions provide the intelligence to pull data from all connected sources.
Three common BI deployment models help aggregate data across the enterprise. In a siloed infrastructure, where reporting is distributed, a reporting server is embedded into each device and sends data to a single console or browser that aggregates all of the input for presentation. In a data center with centralized reporting, multiple databases can feed into a single reporting server, which then distributes reports to a Web browser or printer, depending on the needs of the user. Finally, in a shared reporting model, multiple application servers share a single reporting server for all clients.
Secure Data Access
Security is crucial, as much of the data that requires analysis in an enterprise is confidential. Any BI system implemented should work with the security processes and standards already in place. For ease of use, the system should support single sign-on - meaning that once a user logs in, he can access the various reporting and analysis tools without having to re-enter his credentials along the way. The BI systems should also support whatever security infrastructure is already deployed, including commercial identity management systems and auditing tools. Look for solutions that include an integrated open source security frameworks such as ACEGI (part of the Spring framework), which offers comprehensive authentication access control and supports a variety of existing security infrastructures, such as JAAS, PAM, CAS and Kerberos.
Flexible Data Presentation Options
Another consideration for choosing the right BI solution is the way in which data is presented. The solution should provide you with the flexibility to meet diverse reporting preferences and be able to output data in a variety of formats. For example, a sales rep might need a PDF file of a graph or chart to take to sales presentations. An accounting manager might need the same data in an Excel spreadsheet so he can manipulate it. An executive might want a dashboard view to have performance visibility on the departmental level, such as sales, purchasing or inventory. Still others might want to access data through convenient Web portals while working remotely.
Delivery mechanisms are also important. How will the reports be delivered? Will they be sent out by email, printed and hand delivered or accessed through a Web site? Will you need time stamping capabilities? Can the users running the report interact with it to refine what they are seeing? Who will be receiving reports on a regular basis, and does this need to be automated? A BI solution should be flexible and comprehensive enough to address the wide range of data access methods and presentation formats that people across the enterprise will demand.
Seamless Integration Provides Ease of Use, Usefulness
For a BI system to be truly valuable, it should be able to integrate with your existing infrastructure. What's the use of adding BI to cut costs and make your business run more efficiently if you have to train personnel to use an entirely new system, one that is separate from key business and data management systems?
Although both open source and proprietary solutions exist, adding BI capabilities to your existing infrastructure tends to be much easier if the solution is open source. Open source BI systems are flexible and modular, and work well with a variety of platforms and applications. By contrast, proprietary systems are closed and often act as a completely separate application. This makes it difficult to integrate the BI functionality into your business-critical systems and applications. Also, open-source solutions typically focus on one capability and perform it well, rather than trying to solve multiple problems and not doing a great job at any one thing.
The best possible solution is to embed BI functionality into existing applications, thereby minimizing the training necessary to get personnel up and running with the new BI system. For example, a large educational institution recently leveraged an open source platform to embed reporting functionality into its admissions and enrollment lead management application. The institution was struggling with an inefficient, time-consuming method of combining data from 11 different campuses to create an important sales report called the Admissions Flash Report. This report provided valuable insight into the trends and behaviors of would-be students during the enrolment process. The institution wanted a solution for creating this report that would help the organization reduce costs and save time through automated report generation and distribution of consolidated data from all 11 campuses.
An on-demand reporting solution was embedded in the admission office's Salesforce.com application. The report was accessible through the same database interface that personnel were already using. Users can now access advanced reporting and analysis features from within Salesforce.com - using the same interface and application they use every day.
Since embedding the reporting and analysis functions into its existing infrastructure, the healthcare institution has slashed time spent creating the report from over 50 hours to almost no time. No training was necessary for IT personnel; they simply click on a tab within Salesforce.com to use the new reporting features. Reports are generated automatically, aggregating data from across the organization.
In addition to being embeddable, the BI solution must be able to work with multiple database environments. Often, when a company is acquired, the traditional "data mart" model is being employed; for example, the acquiring company may be running DB2 while the company being acquired is running Oracle. Many BI solutions perform parallel pooling to get around this issue, but such an approach leaves little room for integrated reporting and comprehensive enterprise-wide data analysis. In some cases, one of the companies will change its database environment to accommodate the other company. But a more cost-effective workaround is an open source BI solution that can work with both environments.
To facilitate such ubiquitous integration, standards should be employed:
- Common process standards, such as data quality, and integration with ETL metadata.
- Data acquisition and staging standards, defining formats and content staging.
- Data quality and process standards, such as how quality issues will be addressed.
- Standards for implementing conformed dimensions.
- Standards for fact table builds that can be leveraged across many of the ETL components.
- Common processes and templates.
BI and your Business
Data in the enterprise is expanding exponentially, making traditional methods of reporting and data analysis obsolete. They're simply not efficient enough to keep up with new IT initiatives, which have become more strategic in nature. Traditionally, BI has been linked to IT versus the business user, while business users continue to define metrics and decisions. Still, IT must play a strategic role in directing implementation and architecture at the lowest cost. New BI solutions are helping businesses make effective use of data, but not all of them are created equally. Valuable features such as flexibility to support a variety of IT architectures and the ability to integrate at all levels with existing infrastructures and applications are key to making BI an integral part of your daily business reporting and analysis processes.
Reference:
- Brian Dooley. "BI Trends and Futures." ITWorldCanada.com, September 6, 2005.
Barry Klawans is the CTO of JasperSoft, a leader in commercial open source reporting solutions. He has over 20 years of experience architecting and developing enterprise applications at companies such as Remedy, Sybase and Bidcom. His specialty is making data more comprehensible.
For more information on related topics, visit the following channels:
