In-depth research data in business intelligence
Resource
Don't let valuable data in white and white in a database that cannot be accessed. Use the appropriate storage, business intelligence and .NET technology to provide decision makers to provide information they need. By Rob Ericsson involves technology: Microsoft Business Intelligence Platform, SQL Server, SharePoint Portal Server, Commerce Server
Enterprises collect a lot of data in daily operations; each interaction with the customer or supplier generates data. Enterprises usually spend a large amount of funding to build online transaction (OLTP) and enterprise systems to record the results of these transactions. The data collected by these systems has grown very quickly - they will grow up every two to three years. Unfortunately, a survey conducted by IBM indicates that when companies are doing business decisions, their validates usually account for only 2% to 4% of all of this data. This means that retailers may lose opportunities to improve product sales, and banks may lose valuable customers, manufacturers may produce erroneous products. No matter what industry you are engaged, there is no product that you have already owned will make you spend more time and money, and you may lose a lot of important opportunities.
Business Intelligence (BI) tells how to use the data already have data to build your own competitive advantage. Bi is combined with your business system, which allows you to convert data into information. By using BI tools, decision makers can find the correct data and discover their importance and take action on the results. Data WareHousing Institute (TDWI) describes this powerful closed loop process as a business intelligence cycle. The business intelligence cycle is composed of five steps: Capture, Analysis, Plan, Action (ACT), and Review (see Figure 1). In this article, I will first describe these steps step by step, then tell Microsoft .NET technology provided by Microsoft in the BI field, and finally give you a clear suggestion in the enterprise.
Figure 1. Commercial intelligence cycles collect data to analyze more than just do some queries for your OLTP database. You must design a proper data warehouse and then extract operational data (Operational Data) from this warehouse, or load the operating data into this warehouse. However, even if you have been carefully designed to data warehouses, data from your operating system is usually not as good as you think. In many cases, the most effective way to improve data quality is to modify them in the source (generated enterprise applications). This reflects the importance of supporting the business executive of the BI project, because the better data may mean that there will be some changes. For example, you may need to modify some customer service systems, making your customer service representative more likely to collect precise customer data. If you don't have the support of decision makers, it will be difficult to put these changes. However, this is just a foundation because if your users don't trust these data, they will not use them, your BI solution will be invalid, don't say you waste time and money. TDWI estimates that data will consume $ 611 billion in additional postage, printing and personnel a year. Therefore, with a data collection quality assurance program to ensure accurate, useful data is quite important.
Next, you need to analyze the data. In this step, policymakers use analysis tool to extract information from the collected data. Analytics tools can have a lot of different analytical applications from simple Excel electronic data sheets. In this respect, in order to analyze the data most effectively, do not use a complex tool that exceeds user decision requirements. The complex analysis tool's learning curve is steep, and most users may do not need all functions. Generally only 20% of users need such a tool. These advanced users are often very easy to identify because their work requires a lot of powerful tools, and they usually take the initiative to let others know their needs. Another 80% of users is a general user (Casual USER), they are mainly interested in fast results. If they think that these tools are too difficult or take time, they may not have it. A bad reputation of the BI product is that they eventually become "shelfware" - refers to those purchased, but never needed. Choosing the appropriate tools for your hand is important to analyze the work itself (see the toolbar "Select BI Tool").
In the planning phase, companies will determine what kind of action taken based on information obtained from analysis. Under normal circumstances, the company's policy makers will plan according to a special purpose. However, you can also use some specialized analysis applications to create action plans based on actual data. For example, using a customer relationship management (CRM) application that can determine a Market Dynamic mode, then create a marketing plan to use these new information. The application itself can identify the appropriate market scope for the target and allows decision makers to concentrate on optimal ways to make marketing succeed in this area.
After making a plan, the next step in the BI cycle is to take action. There are many kinds of actions, including realizing a new marketing plan, hiring more telephone salesmen, making them overtime, changing suppliers for a production line. At present, most analytical applications do not have direct integration with enterprise systems that control corporate behavior; combining these applications and operating applications is a clear trend. A good way to combine analytics and operating applications is through enterprise portals, through this portal, users can see the results of the analysis and take action through the same boundaries.
Before the end of the BI cycle, you need to check the results. You can define and collect new metric to help you estimate and manage these results. These metrics, sometimes called key performance indicators (KPI), can help you simulate and track your company's performance. If the performance of the key field does not meet the target, then these metric can help you concentrate on raise performance in these areas. A factory accident indicator may be a simple example of metric. If an accident exceeds a specified limit, you need to evaluate the most recent accident to see if there is a system problem, which can be solved by better training and operational steps.
According to the survey of Survey.com, the global market for data warehouses / BI solutions is expected to grow by 25% between 2000 and 2005. If you take into account financial profits that implement the BI solution, then this data is not surprised. IDC found that the average investment returns (ROI) of commercial analysis applications were 431%, of which 63% were returned within two years. If the large-scale IT project has great complexity and risk, then these results will leave us a deep impression.
Analysis is part of all the work of knowledge workers, everyone needs to access tools and data that supports their decision-making processes. .NET provides some interesting tools for you to choose from, which can help you use data most effectively and use this data to your competitive advantage. Assess .NET technology Microsoft has provided a platform called Microsoft Business Intelligence Platform. One of the cores of the platform, Microsoft Analysis Services provides an OLAP feature for the .NET platform. Analysis Services is a package for SQL Server 2000 (which is part of Microsoft .NET Enterprise Server). Analysis Services is available in the form of CUBE, and CUBE is data organized together with multidimensional structures, which is defined by a set of dimensions and measure. When building CUBE, you first copied data from your enterprise system to a data warehouse or Data Mart (see toolbar "familiar with BI terminology and technology). You can simplify the process with the Data Transformation Services (DTS) or a third-party extraction, conversion, and loading (ETL) tools included in SQL Server 2000. To create a Cube, you need to set a set of points on different dimensions and save data for later reading and analysis. This article provides more details on how to create Cube, "Introduction to SQL Server 2000 Analysis Services: Creating Our First Cube" provides more details (see resource).
Figure 2. Displaying data by Excel Pivottables Analog Pivottables supports three different Cube storage methods, each method has advantages, and it is not enough. MultiDimensional OLAP (MOLAP) saves data and all collections in a separate file. MOLAP provides the best performance, but it also takes up more disk space. Relational OLAP (ROLAP) saves a collection in your data warehouse. This method takes a lot of disk space, but it is slower than MOLAP. Hybrid OLAP (HOLAP) is a combination of MOLAP and ROLAP. It saves a collection as a separate file, while the details are saved in the data warehouse. This approach is less than the disk space occupied by MOLAP, but it is faster than ROLAP, but the collection is slower than MOLAP. Once Cube created, the user can access them through a variety of client applications locally or across the Internet.
Office XP (one of the core components of Microsoft Business Intelligence Platform) is used to access the main client applications for CUBE. Excel is the most commonly used, and you can create Cube dynamic views with Microsoft Pivottable Services (see Figure 2). This allows analysts to access important information with tools they have already familiar with.
You can also use the latest member of the Office family - Data Analyzer - Data Analyzer to display and analyze data in a graphic form, which can provide multi-dimensional enterprise information on one screen (see Figure 3). Data Analyzer allows users to slice the OLAP data, and don't need to be an Excel expert. The simplicity of Data Analyzer and graphical nature make it an ideal tool for decision makers at all levels of enterprises. Other components of the platform, including SharePoint Portal Server, MAPPOINT, and Commerce Server 2002, are all of these core components around Microsoft Bi Platform. Although these components are somewhat unsatisfactory and publishing a universal BI application, they do provide some features that you can use these features in a specific environment.
Figure 3. Discovery trend by Data Analyzer In these three components, SharePoint's application can be the most wide. One major factor in the success of the BI program is whether enterprise decision makers can get useful information in their daily work. It is also important to share and analyze results in decision makers. A good way is to use a SharePoint-based enterprise portal that allows us to send and share BI content and tools in a customized, secure platform. As I mentioned earlier, enterprise portal is also the best location integrating your analysis app and enterprise system (see
Figure 4).
MAPPOINT can appear information about enterprise geospatial information, allowing users to better understand this information. Analysis of geographic locations and demographics can provide new opportunities. For example, if your product is sold very well in a particular user group, the product is positioned to a user base with similar features to improve the targeted market activity more effectively.
Commerce Server 2002 has a Business Data Warehouse, you can use it to save data from online business in order to make further analysis. Commerce Server 2002 analysis features include clickstream applications, purchase, and browse history, product catalog data, promotional activities, and user information (User profile). These features can help you make full use of your e-commerce information, improve sales and profit. For example, if you find the shoppers on your website, you will try to add goods to your cart. Later, when they learned how much it takes to spend how much money, then canceled the purchase activity, then you may be free shipping. Promotions to see if it will bring more profits.
Figure 4. SharePoint integrates analyzing applications and enterprise applications
Create from the beginning
In addition to the components of Microsoft Business Intelligence Platform, .NET Framework itself may also be an important part of your business BI architecture. One of TDWI shows that 62% of companies in companies that have deployed an analytical application are built. Using .NET Framework to build your own unique analysis app is a way to make your business in the competition of our opponents. Because .NET Framework has strong data access components and easy-to-distribute Smart Client, create competitive and useful analysis applications is simpler.
.NET provides the ability to encapsulate existing applications with Web Services to integrate data with internal systems and business partners. Because data access is a key to building the best BI application, .NET can provide a way to convert existing systems into useful components of your BI architecture. You don't have to replace the existing system to get good BI; an alternative, you can use .NET to consolidate the data to better analyze.
In addition to providing traditional BIs, .NET can also provide an opportunity to closely integrate analytics applications into the operating application. Under the rapid economic development of rapid economic development, high operational efficiency (from analyzing the entire process of action is not interrupted) will be a very powerful competitive weapon. Indeed, this integration will be "killer" in the 21st century. However, since this technology is still very new, there are some immature places where the BI application development is developed. One of the aspects is that an ADO MULTIDIMENSIONAL (Adomd) is not implemented in ADO.NET. This means that if you don't have to use some non-.NET code, you can't create a .NET application that uses Adomd. The simplest solution is to use the COM interaction feature in .NET to process the COM Adomd library, but do not use the benefits of hosting code in your entire application. There are additional methods to solve this limitations, but when Microsoft has released the multidimensional functions in ADO.NET, there will be a complete .NET solution.
All in all, .NET platform provides some very attractive technologies that make your BI easier and more effective. Built your BI architecture above Analysis Services, Microsoft Office, and .Net Framework, it will provide more features when the technology is gradually mature.
Implementing a BIBI project in your company is very similar to other IT projects, but there are also some unique complexity you should know. In general, the difficulties to build a BI solution can be reflected in a survey made by David Consulting Group. The survey claims that the data warehouse application is used to measure The rate is much lower than half of the transmission rate of the e-commerce application (see resources). Although this does not mean that the difficulty of creating a BI application is doubled than the difficulty of creating an e-commerce application, it proves that the BI application will be more complicated in terms of analysis, design, build, and transmission. In addition to management of the general fields required for each application, four aspects in the BI project should also pay special attention: metadata, data, underlying architecture, and applications.
The most difficult place to build the first BI project is to collect metadata (describing data for your data). If there is no precise metadata, policy makers don't correctly identify a particular analysis result. However, metadata is always difficult to define, track, and require comprehensive cooperation of major experts.
It is also difficult to design the ETL process. As I mentioned earlier, the quality of data is important, and it is not what you should consider at the end of the project - you should consider it from the beginning. All in all, your quality of your entire BI project depends primarily on your data and metadata quality. It takes some time in this regard.
In essence, the BI application has a lot of related data. Although you can analyze the data, people's processing needs are great. The BI solution makes the user's most headache, the performance is very slow. There are a lot of demand for BI applications, and you need to carefully design their underlying architecture in terms of availability and performance.
Build an analysis application and build other commercial applications very similar. However, because most analytical applications have strong visual performance, and have a broader user base, you need special attention to availability. Therefore, compared with a limited commercial application with a user base, it requires us to spend more time and consider more factors.
There are very few other areas to provide you with so much benefit like a BI solution. However, these benefits are with certain risks. .NET provides some good technologies that enable us to create a good BI underlying architecture in the enterprise, but a successful solution also requires people to work hard to implement it, and carefully plan it. About the author: Rob Ericsson is an experienced software developer and project manager, he is mainly working on Microsoft technologies. His e-mail is dotnet@l10systems.com.
