Introduction
Over the course of the years, computers have become an integral part of the business environment. This has been as a result of their efficiency, accuracy, and speed which has revolutionalized the manner in which business is conducted.
An important concept that has risen with the continued usage of computers is that of Data Warehousing and various architectures of the same have been implemented.
This paper shall summarize and analyze an article by Ariyachandra and Watson “Which Data Warehouse Architecture is Most Successful?” which attempts to provide an empirical research on the strengths of the major warehousing architectures in use today.
Article Summary
Ariyachandra and Watson begin their article by highlighting the fact that there exists huge controversies over which is the best data warehouse architecture. They articulate that the major architectures in existence are the “hub and spoke architecture” and the “dart mart bus architecture”. The authors lament that despite the importance that an architecture choice can have, there has been little research on the topic. As such, this article is intended to shed light as to the different architecture choices.
In their study, the authors primarily investigate 5 data warehousing architectures namely: “independent data marts, data mart bus architecture with linked dimensional data marks, hub and spoke, centralized data warehouse and federated.” The authors noted that while this was not a conclusive list of all the architectures in existence, this sample consisted of the core ones since other architectural types were a variation of the five.
The study is achieved through a Web-based survey which is targeted at individuals who undertake physical data warehouse implementations at various organizations. The surveyed companies were small to large companies. From the study, it was discovered that the mostly popular architecture was the hub and spoke followed by the bus architecture. Since this data warehouses had to be used in conjunction with a platform, it was observed that the most common platform was Oracle followed by Microsoft.
Since there had to be a measure by which the relative success of a particular architecture was measured, the authors chose to use the following attributes: “information quality, system quality, individual impacts and organizational impacts.” By contrasting the score that each architecture acquired on this varied matrixes, the authors were able to compare the success of the different architectures.
The findings indicated that Independent data marts were the least popular as they received the lowest score on all counts. The authors conceded that this was in line with the commonly held perception that an architecture that favored the independent storage of data was undesirable by most organization.
The studies furthermore revealed that the 3 centralized architectures only had marginal differences in their average score. This emphasized the popularity with which centralized systems of data storage were viewed in the industry. From these findings, the authors declare that none of the three architectures can claim superiority over the others.
In conclusion, Ariyachandra and Watson acknowledge that the reason why the competing architectures have survived without any one of them gaining a significant edge over the other is because they are both successful for their intended purposes.
As such, organizations are content to utilize any of these three architectures since no single architecture is dominant in system quality of informational quality. The authors also note that as the differing architectures have evolved over time, they have become more similar in nature which explains their almost similar scores on the success metrics.
Analysis of the Article
Main Issues addressed
The article underscores the importance of data warehouses to any organizations success. A data warehouse is a center of information and its administration and it is considered as a “subject oriented, integrated, nonvolatile and time variant collection of data in support of management’s decisions” (Blazewicz 393).
As such, the article asserts that making the right architecture choice for an organization’s data warehouse is an issue of great importance. While there exist numerous studies that describe the various architectures, the authors of this article illustrate that there is little to no empirical research through which the success rates of the various architectures can be compared.
The study that the authors undertake to fill this informational void that is created by a lack of studies dealing with data warehousing architectures reveals some important truths about the various architecture types in existence. The main discovery is that independent data marts are hugely unfavorable since they do not allow for interactive accessibility of information to assist in decision making.
Relationship of Article to the Course
The authors highlight at the start of the article that its major purpose is to investigate the successes of the various data warehousing architectures. The article places emphasis on the importance of a centralized database system, a fact that has been articulated in this course.
From the article, it is clear that at an industrial implemental level, centralized architectures are preferred to independent data architectures. This is because a data warehouse consists of a variety of different data that is collected, unified and updated and placed in a manner that is accessible for management support in decision making. In such a scenario, having a centralized integrated database is more useful than having stand alone databases.
My Thoughts on the Article
I found the article highly useful as it gave a brief yet informative overview of the various data warehousing architectures and their success rates in the industry. In my opinion, the authors wrote the article in a clear and precise way which was highly readable.
They avoided overloading the reader with a lot of background information on the topic and instead concentrated on giving a discussion on their study, its findings and the significance of their findings. A figure was also utilized to show the five data warehouse architectures that were studied. By using this figure, the reader could at a glance tell spot the significant differences between each implementation. This was therefore a very effective mode by which information was provided in the article.
The findings of the study which had been conducted by the authors were provided in a tabular manner. Not only was this a highly readable means by which to display the results but it also enabled one to compare the results for the different data warehouse architectures at a glance.
The authors also reconciled their studies with the commonly held perspectives that centralized architectures were superior to independent ones. The study conducted included a targeted web-based survey. Since the survey resulted in 450 respondents giving useful information, it can be said that this was a big enough data set from which the survey results could be obtained.
The Survey also covered a wide range of companies (small to large companies) hence giving a plausible picture of what was happening in the industry. Various measures were also used to assess the success of a particular architecture in the study. This ensured that the completeness of an architecture was also taken into consideration since if an architecture fulfills only one measure and fails in the other, it cannot be said to be ideal.
However, the length and scope of the article was not appropriate given the depth of the subject that the article as discussing. While the concise manner in which the article was written might appeal to people who have a working knowledge of data warehouses, it is hugely undesirable to the non-expert reader.
This is because the authors do not even offer brief descriptions of concepts such as “data warehousing” which are central to the article. This means that the article would be of little use to a person who did not possess at least a basic understanding of the subject. It is therefore obvious that the intended audience for the article is the technically savvy group which had an in-depth knowledge of the subject matter
Conclusion
This paper set out to summarize and perform a critical analysis of the article “Which Data Warehouse Architecture is Most Successful?” By Ariyachandra and Watson. The article gives a report of an empirical research in which the strengths of the major warehouse architectures are measured.
The contribution of this article is immense since a comparison of the major architectures has not being undertaken by other previous studies. However, the brevity of the article renders it unusable for the novice reader. Its appeal to an expert audience is however great and the information presented therein invaluable.
Works Cited
Ariyachandra Thilini and Watson Hugh. Which Data Warehouse Architecture Is Most Successful? Business Intelligence Journal, Vol 11, No. 1.
Błażewicz, Jacek. Handbook on data management in information systems. Birkhäuser, 2003.