Part one of this article discussed the difference between traditional BI and embedded/operational BI and some of the business considerations to identify when looking to implement BI. Part 2 explores the technical and business considerations to take into account within both business and IT departments.
The key differences between traditional BI and embedded or operational BI is the way it is applied within the organization. The general structure of BI is the same, with data being captured and analyzed more often to enable decision making on a regular basis using daily data updates. These slight differences may affect an organization’s IT infrastructure. Therefore, organizations should identify potential technical considerations to discover the best way to manage BI within the organization while taking into account future expansions of both applications as well as users within the organization.
Technical considerations for BI differences – between traditional and OBI
Once an organization has a general understanding of BI’s main components and how it works, it becomes possible to decipher what technical differences exist based on the deployment of BI when general architecture may be similar or even the same. This means that although the same infrastructure may exist, the technical considerations required to adequately maintain a BI infrastructure can differ based on the way business intelligence is deployed within the organization. For the purposes of this article, extract transform and load (ETL), server space, and the frequency of transactions will be discussed.
ETL
The first step to delivering valuable results through business intelligence is to identify the data needed to make informed decisions. By looking at the business problem and working backwards to see which data is required to provide answers to business pains or to current gaps in knowledge, organizations can identify the data sources required. Addressing business issues might require a subset of information from five or ten different data sources. Within ETL processes, this information can be identified and loaded into the data warehouse in the form required for analysis. This means that raw operational data is transformed into an analytical tool that when added to front end analytical applications and reports can be used as an aid to solve the business issues affecting the organization.
Generally, organizations should be aware of more than just what information they require. In some cases, information used in general reporting is consolidated, have algorithms associated with output, etc. that make it almost impossible to create duplicate output without knowing the intricacies of how the data interrelates and how the original information was created. One example regularly used to illustrate this problem is when executives have planning meetings but the information they are coming to the table with differs from one another and creates a gap between what each executive sees as the organization’s reality. This example highlights the importance of having a resource that is familiar with the business rules associated with the data itself. Once information is added to the data warehouse a centralized source of reporting and analytics can be distributed to executives and other decision makers within the organization.
The freedom of BI enables end users to slice and dice data to create valuable information views and helps with the decision making process. However, if different versions of reports still exist within the organization, it becomes more difficult to identify which version is the most accurate.
Overall, ETL processes enable organizations to identify, to capture and to load relevant data into the BI infrastructure to use for more in-depth analysis. Organizations should spend the appropriate amount of time identifying the right data sources, what subset of information they require and the additional transformation requirements, such as data cleansing and ongoing data quality initiatives. Because data integration activities may encompass the bulk of initial BI efforts, strong processes should be put in place to ensure the ability to maintain as well as to change data input. In addition, as technologies evolve, the ability to access operational data stores and to interact with the data warehouse creates a different dynamic regarding how data is captured and managed. Consequently, as ETL needs are being identified, the various ways in which organizations can interact with their data should be identified to consider the most appropriate data capture and data management for the organization.
Servers and the amount of data
Due to large amounts of data required for reporting and analysis organizations should identify the current and future resources needed to compensate for the additional data captured and stored to enable BI. A great benefit of business intelligence applications is the ability to house historical data to analyze trends and to help plan for the future based on past performance and the success of previous initiatives. Consequently, the amount of data stored within a data warehouse can be exponential, especially when considering intra-daily data updates.
This means that the required server space and overall performance might become an issue if BI is not planned correctly. Within the traditional BI approach, data is pulled into a data warehouse and may be aggregated, calculations performed, etc. to bring information to decision makers that can be analyzed within the overall context of their business. Therefore, subsets of data need to be loaded on a regular basis and retained to identify trends over time. Because of the large amounts of data that are stored, appropriate space and resources should be allocated to manage these processes and to take into account the potential for growth and future additions to both data and end user requirements. For instance, although BI may initially be implemented to solve an issue regarding sales performance, its use may extent into employee performance or marketing. These additional uses may or may not require new or extra data capture. Even if this is not the case, the addition of end users can also tax an already high-volume data traffic environment.
Also, depending on how information is aggregated and how data is analyzed, storage considerations will differ. In some cases, data is aggregated and algorithms identified within the data warehouse or data mart. In other cases, these calculations occur in memory once a request has been made (i.e. sales results by region or by product) relieving the burden of having to store calculations along with data. Either way, the infrastructure needs to take into account when and how BI is delivered and whether these areas will affect the deliverables required to deliver analytical applications to the business timely and efficiently.
Right-time BI
When BI is delivered can become as important as the data captured. The pattern of updating data and delivering intra-day data updates within BI is one of the requirements of embedded/operational BI. Depending on how an organization conducts decision-making practices and what problem BI is helping address, the number of times data is refreshed becomes important. Consequently, right-time BI – the ability to have an accurate and timely view of what is happening within the organization – becomes an additional consideration on top of both ETL and server space requirements as the three become interconnected. Realistically, all three are interrelated as the ability to deliver the right data at the right time to the right people has become one of the selling points of using BI to help with an organization’s decision support.
The ability to provide regular updates to the data warehouse enables BI to be deployed to more decision makers throughout the organization. In many cases, executives only require a snapshot of how the organization is performing, but line of business (LOB) managers and employees need a constant view of what is happening within the organization to perform their daily tasks more efficiently.
Deploying right-time BI may not be appropriate for every organization as different organizations use BI to solve various business issues, to manage metrics, and to measure performance. Strategic decision-making requires a bird’s eye view of information, meaning that intra-daily updates are not beneficial. Alternatively, within operations, seeing up-to-date information to identify employee performance and service levels become essential as an organization’s focus on quality increases.
What data considerations mean for business
For any business application to benefit the organization, the focus has to remain on the use of IT to help solve business pains and to enable a broader view of performance. This focus enables organizations to stay competitive within their respective markets. As business intelligence becomes more pervasive within organizations and as technologies advance, the way in which BI is deployed will change – as will the general technical considerations. What this means is that organizations should take into account that their BI applications require the flexibility necessary to help the organization keep pace within a constantly changing environment that focuses on the increase of end user interaction and autonomy. By taking into account technical considerations such as where the required data is stored, how often data refreshes are required, and how to house that data to ensure timely delivery, organizations can ensure that business requirements are being met by using IT to help drive those business decisions.
Copyright 2008 - Dashboard Insight - All rights reserved.