Open Source Solutions

As we prepare for our talk at Campus Technology 2007, we're considering some of the topics we might cover.

First off... "What is BI?" Even on WikiPedia, this is a topic of conversation, as you can see by clicking on the "discussion" tab of the Business Intelligence article. For our purposes, we'll define business intelligence as business processes and the integrated applications or individual tools that are implemented specifically to provide data management, reporting and analytics to solve specific business and user community needs.

As with other IT initiatives, a BI program can be built from scratch, built from an existing framework, such as the open source Eclipse BIRT or the proprietary tools from SAS, purchased as a COTS product, bought as SaaS, obtained as an appliance or implemented from Open Source projects.

I can't imagine an IT shop that hasn't faced the decision to build or buy, and then to decide how to build or what to buy. And many IT shops have developed a culture that leans one way or another. Today, however, there are some new wrinkles... new variables in the decision making equation. There are more options than ever before:

• more language options,
• more vendors and more vendor consolidation
• more libraries,
• more open standards,
• more architectures,
• more services, and
• more concerns.

These options provide more flexibility in meeting organizational and user needs, but more challenges in leading the way to a responsive, secure, compliant, cost effective and maintainable information infrastructure.

Some of these options are fairly new, and can have wide-ranging impact for internal standards and procedures going forward. Among these new options, especially in BI strategies, are Appliances, SaaS [not SAS] and Open Source Solutions. Even more confusing, some appliances use open source software components, some are strictly proprietary in all their parts, and some are a mix; while some open source vendors make their software available only under an approved open source license, and others use dual licenses: open source and proprietary, sometimes with different features or different names for the different licensed versions. Oh, and we're not even going to touch the myriad of open source licenses out there, nor the debate over whether or not certain licenses or certain vendors are "really truly open source".

Every situation is different, and we're not going to try to solve all the world's problems, or even BI strategies in a simple blog post.

One point that I would like to make is that BI is, IMNSHO, especially well suited to open source solutions. At their best, proprietary solutions still only give one a starting point, which must be customized for the source system customizations and uniqueness inherent in each IT shop, as well as the specific business processes, organizational considerations and user desires that are driving the BI program. Here are some examples of what I mean.

• A BI solution is likely to take from a variety of source systems [financial, human resources, full ERP, grant management, housing, student records, large legacy software, new enterprise applications, small local databases and spreadsheets], and may have different real-time and historical requirements than those source systems. You may be considering implementing SOA throughout your enterprise, and exposing some or all of those potential source systems as services. Thus, do you connect directly or do you need ETL, or maybe a mix of ETL and ESB, to get your source data to your BI system? Whether you purchase a proprietary system from Informatica or Oracle or Data Stage or Tibco or BEA or IBM, or download an open source solution such as KETTLE or KETL or Jetstream or Mule or ServiceMix, you can't just install the software, auto-magically connect to all of your source systems and have a nice working system at the end of the day.
• Any reporting tool just gives you the ability to design and format reports. There probably aren't any nice canned reports, out of the box, that will make your users dance in the hallways. So, do you buy Crystal Reports, or do you download JasperReports, jFreeReports, or openReports?
• The same can be said and asked for analytical tools such as OLAP and MDDB, predictive tools, data mining and dashboards.

The point is that even for small organizations, or simple reporting needs, the best that you get from proprietary vendors are the tools to create the BI solutions needed, and the best that you get from open source solutions is exactly the same, plus the actual source code to customize if need be and a community to help share the pain. Proprietary BI solutions can cost hundreds of thousands to millions of dollars or euros or marks or credits in licensing costs plus consultants, plus programmers, plus hardware, plus support. Open source is much the same on the plus cost, but greatly reduced in the licensing cost, sometimes to zero. The bonus may be slicker wizards to help your programmer-analysts deliver the solution in a proprietary vs. open source case, or greater flexibility and control in the open source vs. proprietary case. The trade off here is the same as those old buy IBM or build COBOL solutions we once considered: what will get the project delivered on-time, within-budget and meeting the specifications?

Again, we're not trying to present the ultimate solution here, but hopefully, we've opened up the possibilities. So, take the lead, and generate your strategy one step at a time. I have to run, but I'll come back later to fill in some links. For open source projects and vendors, take a look at our linkblog in the side column.