Table of Contents
1. INTRODUCTION
2. THE
MAIN ISSUES CONCERNING DATA WAREHOUSING
3. CASE STUDIES
3.1 LEXMARK INTERNATIONAL
3.2 ALLTEL
4. SUMMARY
5. RECOMMENDATIONS
6. REFERENCES
1. INTRODUCTION
Data is at the heart of corporate survival. Consistent and valid data is needed to define and continually reassess corporate strategy. (Ferdinandi, 1999). However, in order to be complete, consistent and accessible in a timely manner, data must be kept in what is currently a hotly debated issue, a data warehouse. As O' Brien (1999) puts it "a data warehouse stores data from current and previous years that has been extracted from the various operational databases of an organisation. It is a central source of data that has been screened, edited standardized and integrated so it can be used by managers and other end user professionals for a variety of forms of business analysis, market research, and decision support".
Nowadays, more and more companies have already implemented or consider implementing a data warehousing project in order to create or maintain the competitive advantage. The aim of this study is to present the cases of two companies that have implemented a data warehouse and evaluate their relative success based upon actual results. The two companies under consideration are Lexmark International and ALLTEL, both operating in the IT-Telecommunications sector.
The reasons behind the choice of these particular companies are two. First, the fact that both companies operate in the IT-Telecommunications sector ensures some comparability. Second, ALLTEL, unlike Lexmark failed to bring the desired results in its first attempt to implement a data warehousing project.
[ Κορυφή ]2. THE MAIN ISSUES CONCERNING DATA WAREHOUSING
Some of the main issues concerning date warehousing that will be under consideration in this study include:
One of the key issues in the data warehousing debate is what is the problem situation that reinforces a company to turn to the data warehouse solution.
What will be also examined in this study, is how the companies under discussion are addressing the issue of whether they decide to go on their own or get help from outside consulting and software/hardware firms.
A hotly debated issue in the data warehouse literature is the choice between an enterprise wide data warehouse or a special purposed data mart.
Another matter of concern is the choice of the appropriate technology upon which the data warehouse is going to be built upon.
Last but not least, the issue of the managing process and evaluation of the results brought from the data warehouse implementation as a basis for future decisions and necessary improvements will be discussed.
[ Κορυφή ]
3. CASE STUDIES
The two cases illustrated below refer to Lexmark International and ALLTEL, both USA-based companies, operating in the rapidly growing IT-Telecommunications sector.
3.1 LEXMARK INTERNATIONAL
The company
Lexmark International is a fast growing global company, which develops and manufactures laser and inkjet printers and associated consumable supplies for the office and home markets. (Reitzes, 1999).
The problem
Lexmark surveyed its reporting infrastructure for measuring order shipment, delivery performance, managing product sales and inventory information. The survey revealed that this method was highly transactional in nature and unable to meet the needs of the company for efficiency and convenience to its customers.
The solving process
Lexmark decided to ask for outside help rather than go on its own, since data warehousing was new to the company's management. An IBM team was brought to introduce the data warehousing concept, present a summary of potential projects and validate the investment. In the same time the assistance of a "quick strike" team of consultants from software developer MicroStrategy was used.
The solution
The company decided to go for a data warehouse. The solution was the customer delivery performance project, which was completed in late 1997.
The technology
The Lexmark data warehouse was built on a variety of MicroStrategy products. It uses Seagate Crystal Reports for query and reporting, runs on a DB2/6000 and captures data sources from J.D. Edwards' ERP systems and EDI transactions. It is designed to store and analyse data related to the life of a sales order - from order entry, preparation for shipment, to delivery to the customer - and thus the customer delivery performance project has automated the entire process of collecting sales and product delivery metrics.
The results
The customer delivery performance project proved to be successful and here are some of
the figures that justify this success:
Re-evaluation for future decisions The success of the pilot data warehousing project led Lexmark to initiate the
development of a second data warehouse, the retail management system pilot project, which
was completed in January 1998. This data warehouse was developed to improve the timeliness
and accuracy of sales information, which in turn has improved Lexmark's ability to measure
the performance of field marketing representatives, develop accurate forecasts, evaluate
the effectiveness of advertisement campaigns and identify market opportunities. This
project was also successful but the report of further details is beyond the scope of this
study
Product delivery cycle time: Customer order placement
to actual delivery to his/her door has been increased by 70%.
3.2 ALLTEL
The company
ALLTEL is a customer-focused information technology company that provides wire line and wireless communications to more than 7 million customers in 24 states in the USA and has information services clients in more than 50 countries and territories. (Erickson, 1999).
The problem
ALLTEL had identified three main problems that needed immediate action:
First of all, the high cost of ad hoc reporting done on a third-party billing vendor's mainframe, and second the high report turnaround time where requestors had to submit their queries to dedicated staff. The third problem had to do with the need for analysis of customer calling patterns, churn and easy identification of the profitable and non-profitable customers.
The solving process
ALLTEL, like Lexmark hired consultants to do the data warehouse design. They interviewed between 70 and 100 end users of the current ad hoc reporting process. From these interviews came a requirement matrix which the consulting team translated into a data warehouse design.
The solution
The company went for the implementation of an enterprise wide data warehouse in order to resolve the problems it faced.
The technology
Many products were evaluated for all stages of the data warehousing process. Finally, an HO T500 was the machine selected. Also the company purchased the PRISM Warehouse Manager to generate COBOL source code and target SQL code as well as to serve as a meta data management tool. BusinessObjectives was selected as the desktop DSS tool, and a traditional RDBMS was chosen for the data warehouse DBMS which was then constructed in normalised form.
The results
From the start of design it took the company six months to get a successful load of their operational data into the data warehouse. In addition to that, it took ALLTEL another six months to get change control procedures in place to manage their loads. The whole project was far from being successful:
Low performance: There were 17 loads per month for two million customers, each taking four to six hours to load summary billing and usage information.
Delivery drawbacks: The company had to scale back the number of tables that were to be delivered in order to get anything delivered at all.
Reduced quality: Quality assurance testing was moved to the end of the process, instead of being done from the beginning.
Poor ad hoc query performance: Although, BusinessObjects desktop tool was found relatively satisfactory by end users, ad hoc query performance was totally unacceptable, ranging from an hour to "it never came back".
End user's dissatisfaction: End users were disappointed with the product and, therefore, preferred to use the mainframe ad hoc reporting process.
Possible causes of failure
The main possible causes of failure identified are:
Re-evaluation for future decisions
ALLTEL managed to analyse the cause of the problem and to learn from its mistakes. The result was the production of an effective data warehouse in four months. This time ALLTEL used its own people, who had experience with the current ad hoc environment and the current DBMS system. ALLTEL discovered that it was the normalised data warehousing design that was the major flaw and thus it was decided the design staff to go to a dimensional modelling class. Informix Red Brick Warehouse was purchased so that the company could move toward a "data warehouse centric" DBMS that was simpler to manage. Also, the PRISM Warehouse Manager was replaced with transformational and load utilities coded by the company's billing vendor.
Since the re-invention of ALLTEL's data warehouse, loads were taking 30 minutes instead of six hours and this allowed the company to add 30 daily loads to the 17 monthly loads. Also, a lower level of detail for billing was added and this increased the size of that data 10 times. Finally, most of the ad hoc queries ran in minutes instead of hours and even the most complex queries were competed in a reasonable amount of time.
[ Κορυφή ]
4. SUMMARY
Data warehousing aims to facilitate decision making by the management and help the business in creating/maintaining the competitive advantage. In most cases successful data warehousing projects with outstanding results come into light, but failures, although not so popular, are not rare. Still, companies that failed in the implementation of a data warehouse can benefit from their experience, by identifying what went wrong and what should be done instead. This was the case for ALLTEL which found out that its first attempt to implement a data warehouse did not fulfil its expectations. However, the problems were analysed, a new procedure was followed, new technology was adopted and as a result a new effective data warehouse was developed.
On the other hand Lexmark International developed its first data warehousing project with proven success. The customer delivery performance project resulted in cost reductions and improved efficiency and this results led the company to initiate a second data warehousing project.
What it can be said for conclusion is that there are not standard procedures or paths that lead to successful data warehousing projects but there are some issues that the management of the business should reflect upon when considering to develop a data warehouse.
Cases derived from: http://www.dmreview.com