No one ever said building a business intelligence architecture is easy, but as data sources multiply and the market becomes downright saturated with BI tools and platforms, figuring out how to do so strategically has become even more complicated.
Especially when a large vendor or a management consultant presents potential customers with an overly simplified
Components of Forrester's BI reference architecture
- Data sources
- Data rationalization, federation, movement, virtualization
- BI on BI
- Integrated metadata
- Embedded BI
- Direct data access (including streaming data)
- Derived data sources
- BI/DW out of the box
- Information lifecycle management (ILM)
- Enterprise content management (ECM)
- Big data, including Hadoop
- Analytical data virtualization
- Data usage
Source: Forrester's Craft Your Future State BI Reference Architecture
"When unsuspecting clients look at it, they'll think, 'All I have to do is buy your product, engage you to deliver services and mine will look like what you're showing,'" said Boris Evelson, principal analyst specializing in BI for Forrester Research Inc. in Cambridge, Mass.
While BI tends to be a topic with a lot of gray, this story is black and white: Large enterprise BI cannot be simple. Its architecture needs to take into consideration an array of functions like different data sources, data integration, data quality, data warehousing and, of course, what practitioners tend to think of when they think of BI, reports and dashboards. Helping businesses traverse that expansive terrain -- and how all of the pieces fit together -- was a driving reason behind why Evelson and fellow Forrester analyst Noel Yuhanna recently released Craft Your Future State BI Reference Architecture, part of Forrester's playbook series.
The new document is meant to act as a guide for businesses building up or building out their BI architecture. Rather than a detailed perspective of actual modules or an endorsement of products, application development and delivery professionals should expect a high-level BI architecture template, culled, in part, from Evelson's more than 25 years of expertise and personal experience.
"I talk to an average of 4-, 5-, 600 clients every year. I talk to hundreds of vendors every year," he said. "Throughout all of those conversations, whatever they have as their best or their worst practices, I see that on a daily basis."
For beginners and experts
The guide was designed for two use cases, according to Evelson. For businesses that have already begun building, it can help reflect or expand on specific components. The data rationalization section, for example, breaks down the different functions needed to work data over so it meets data quality, master data management and data governance standards, essentially preparing it for staging and analysis.
As businesses go through and investigate how they can improve their architecture, overlaying what they've deployed with the reference guide, they can identify specific components they don't have, and find gaps in the components they do have, Evelson said.
But, he continued, the reference guide could also "introduce them to new opportunities to do something better, faster [and] cheaper, that they haven't thought about before." For businesses interested in analyzing semi-structured data, a quick glance at the guide reveals text and natural language processing (NLP) tools should also reside in the data rationalization section.
More on BI architecture
Find out how Hadoop and open source BI software impact BI architecture
Learn the seven aspects of a successful BI architecture
Consider how to expand your BI architecture to mobile devices
Second, for those businesses starting from scratch, the guide can provide a tutorial, breaking down individual components and explaining how they fit together. This can come in handy when creating a road map for their BI architecture, and it can enable them to "start and proceed small, but think big," according to the report.
That doesn't mean businesses will need to invest in every single tool or component they find within the reference architecture, Evelson said. "Some functionality will definitely come as a package," he said. "But what you need to do is make sure you understand what these components are. These are the functions that -- somewhere, somehow -- need to be performed."
That level of comprehension can be a huge win for businesses, especially as they continue to experiment with processing data in the cloud. The reference guide acknowledges the trend, but cautions that the cloud should not be an excuse for a lack of expertise. Businesses that know what functionalities to expect from their BI architecture as a whole can arm themselves differently from those that don't. Some cloud vendors may, for example, assume businesses are handing over clean data when they're not, Evelson said.
"That would be a gap," he said. "And you would know that even though you're going to the cloud with this, and a cloud vendor will build your dashboards for you, you'll still need to do your data cleansing."