1023-1025

Previous Table of Contents Next

Page 1023

summarizes only by region. You then can quickly query the values in the workbook and click on a graphical object to show your total sales for each region.

If you need to, you can change the fonts or the labeling of this graphic. You can even capture it for use in an online file or an HTML file. Discoverer 3.0 is a powerful OLAP tool that can generate Web images for a corporate OLAP intranet.

Suppose that you decide to make your initial report a bit more professional looking. You can create break areas by using the Sort option on the Tools menu of the User Edition. You also can right-click the column headings and make them bold. Then, you might want to click each column to select it and the data below and give each area a different background color by using the brush at the top right of the menu. Finally, you might select the Dollar Sales SUM column and give it a dollar sign to make the report look more realistic. The output of this summation report has multiple breaks in it, ending with a grand total for sales.

TIPS FOR WORKING WITH THE USER EDITION

• Analysts should work closely with the Discoverer 3.0 administrator. Many join, hierarchy, and folder relationships yield a completely different type of report. If certain aggregates are not held in a data cube (a summary folder), they will have to be computed at runtime.
• Analysts need to understand the relationships set up by the administrator so that they will understand the limits to drilling up, down, and sideways to other folders.
• Data must be refreshed in the Discoverer 3.0 OLAP database from the OLTP or data warehouse within a realistic time frame. If the update of the data cubes occurs only once a month, the data will not gain the trust of the daily users of the system. If the data is updated every 10 minutes, the process could severely impact performance on both the production system and Discoverer 3.0.
• Analysts should be familiar with other graphical and publishing tools. Because you can put graphics on a clipboard or send them to Excel, much of the final product with the User Edition will be integrated with more traditional publishing and presentation tools.
• Use a practice database, such as the one you created in this chapter, to train analysts. If you have a large data warehouse, it will be a slow and CPU- intensive process to train users on the aggregation and drill-down functions of the User Edition using your data.
• Use query- governing tools. Oracle enables the administrator to limit the length of long-running queries. If you have set up your summary folders correctly in the Administrative Edition, most queries shouldn't take more than a few minutes. Long-running queries tie up resources and slow down other users. Also, a long-running query could fill up the extents of your sort tablespace that is being used for OLAP

   continued 

Page 1024

 

 continued 

• users. (Extents are placeholders Oracle uses when data is stored.) Therefore, if one user fills this tablespace, all users receive the error Cannot create extent. One way around this problem is to create a very large temporary tablespace for a user called BIG_QUERY. If users need to run intensive queries, they can log on as this specific user.

• Communicate constantly with Oracle. For a new product of such high technology as Discoverer 3.0, Oracle will constantly release more efficient versions regarding the way it handles data in the star schema, the way in which relationships are described and reported against, and the overall flexibility of the tool. You should never buy this tool without OracleMetals support and a CSI number.

Multidimensional Databases: The Possibilities
Are Endless, Spock!

The use of multidimensional databases is only in its infancy. The storage mechanisms of the data cube will improve over time, as will the interface to relational databases. After Discoverer 3.0 and other products like it improve, what are the next steps?

• Data mining: Right now, Discoverer's User Edition enables you to graphically display and report patterns in your data. With previous Oracle tools, you probably couldn't see or at least format such patterns without a tedious effort. This capability will help business analysts visualize and study these patterns. The next step, of course, is to let the computer find the patterns in the data by using data-mining techniques to perform statistical analyses on the data. Many times, patterns you see are not significant, or there may be a very significant pattern that is hidden by the different dimensions of the data. Using automated data-mining tools enables you to see relationships in your business and determine whether they are significant.
  Data mining needs the multidimensional paradigm as much as the paradigm needs data mining. I once knew a data-mining addict who craved for new sets of data to harvest. After a while, I gave in and obtained data from a client of mine. For two days, he was stuck in a simple DOS editor flipping the dimensions of my data. Finally, on the third day, I heard a sound of ecstasy; he had transformed the data into a multidimensional format that finally was being mined. I think he could have used a multidimensional database!
• Real-time reactions : Suppose that you are the head buyer at a department store around Christmas time. In the toy department, Captain YoYodine begins to outsell all other toys. If you have a very quick refresh rate between your OLAP Discoverer 3.0 system and the online sales system, you can sit in a control room, view the changing patterns

Page 1025

• in the data cubes, and make quick orders from your warehouses to maximize a rush on certain items. By monitoring very fast-moving businesses, you can see patterns more quickly and adjust to the market faster than your competitors .

• Data warehousing: With Discoverer 3.0, data warehousing becomes much simpler. In a typical data warehouse, you have nightly, weekly, or monthly feeds from a series of online systems. You also have a series of batch processes that roll up this information into summary tables. All these interfaces, data structures, and programs greatly increase the development time and cost of a data warehousing project. Why reinvent the wheel?
  With Discoverer 3.0, you can eliminate any rollup tables needed by a data warehouse, because you are storing summary data in a data cube or, as they call it, a summary folder. Also, Discoverer 3.0 makes it easier to convert columns from disparate online systems into simple business concepts. With this OLAP tool, you can save on resource and labor costs to build and maintain these automatic processes. In the future, a data warehouse will be a common type of database ”not one that just wealthy organizations can undertake.
  For more information on this complex topic, see Chapter 58, "Data Warehouses and Data Marts."
• Web publishing and intranets : With the precalculated data cubes of an OLAP product, you suddenly have a server with valuable business information residing on it. A company can create a uniform interface to this data by creating an intranet that will allow many users on different systems in different locations to log on using the standardized Web browsers and protocols (HTTP) of the late 1990s. Instead of building their own separate systems, departments now can take subsets of the data cubes that relate to their business area. This capability eliminates the need for different branches and locations of a company to reinvent the wheel with similar data-analysis tools.
  Aside from the intranet possibilities, OLAP can allow users to log on and view data over the World Wide Web. Because of its inherently graphical nature, it is a perfect tool to display data on the Web. For example, valuable economic data could be online on a Web site directly connected to a multidimensional database and OLAP system. Providers could charge subscribers for access to a certain type of online data summary.
  For example, suppose that you create online feeds into an OLAP system from the stock market and related financial indexes. If you could prove that your multidimensional database isolates important factors that increase profitability for traders and other buyers of financial instruments, you could charge basically anything for access to your online OLAP Web-based system.
• Virtual reality: With the advent of Virtual Reality Modeling Language (VRML) as the standard Web language to handle 3D objects, you can realize the future of William Gibson and his Neuromancer. With a 3D language such as VRML, bankers of the

Previous Table of Contents Next