Data Warehousing - Tom \ 's Ten tips

Data Warehouse is a new 90 \ 's, who has promised to provide data for the landscape well. When we get there? Many providers of market, because it is desirable that data throughout the organization, and it is always useful in the future. What is the size of the data market to grow, it is not known yet. But it is still rapid growth, and is currently at 4.5 billion dollars per year (IDC). 1. Why Data Warehouse projects as regards Creep? Citation Bill Inmon (guru and author of several books on data warehousing) \ "with the traditional and projects with the data. Data Warehousing projects with data and ends with the conditions." Once the project is on track, users find new applications and, consequently, new requirements for data transmission. It is interesting to note that these projects are often justified, R & R work away from \ 'homme de données \ ". What we \' I saw you first, what is happening, once the project requires the questions are more on the applications of these \ 'homme data ". This message can be displayed for the first case, the companies, but in fact signals the beginning of the value added of the project DWH. 2nd Star Schema Model Versus Magnitude Relationships? He has done great in the debate on the merits of the various data models. At the risk of over simplification: ER-models generally have better performance (conversion) to the final consumer, and are often regarded as easy to understand for end users. Disadvantages are that the ER-models need more space, and because of the inherent redundancy of data, problems of consistency from a maintenance perspective. However, the practice seems to be that often a combination of the two is inevitable in the practice of recruitment, despite the parameters (ER or Star) of the head architect. Total Star models seem to have won most soils. 3rd The importance of a Data Warehouse Business Case Much has been written about the business case in a Data Warehouse. What happens when a good deal? IT-savings throughout DWH Business Cases. The important point is not only the "pure" savings, but a connection to the main job, as far as possible. For example, the cycles of faster recovery of the list are selected (quantify if the hourly rates), but is even better if the revenue from acquisitions more customers, resulting from this choice may be linked not only with respect to sales growth of economies, rather than an economic equilibrium, the most important is inherent to the company buy-in that the results of direct access to the company bottom line. Today, changes in legislation (in particular the Sarbanes-Oxley) play an important role in the justification of commercial opportunities. Whether by an improved assessment of companies to gather information transparent, or less for the night CEO, is not ... 4th Why Data Warehouse Projects \ "Never \ 'go wrong? Indeed, plans for data warehouse and sometimes not. But it is not so rare that it is very hard to believe ... In particular, after so much anger for the end user. And there are many opportunities, such as a data warehouse project can not go. Concrete over time, administrative data and data quality inevitably feed. Policy (see Tip # 7) are probably the best explanation for this phenomenon, almost 100% success on projects DWH. In my experience, is the reason for failure or "semi-error" can not be ignored, because the level of higher management, it is not known or, let \ 's say "disheartened" to speak of the Fund misspending. This resulted in not enough. There may be as the consultants have an interest in it, with, as it provides the industry a little 'more activities ... J 5th What is the difference of Web Data Warehouse? Kimball & Merz (2000): "Although these data paths, which in many cases it is raw and unvarnished, which has the potential, who have never known the details of every gesture, every man with the Internet media." The type of data with unique challenges subatomic path. There is less feedback mechanisms to ensure data quality, compared to other data streams. The relationship between the click of a mouse, the user, and the voices of server logs is not as dense as in "traditional" transaction processing due to technical problems, such as proxy server and cache. Because of these differences, staff need to adapt to the end of the web, rather than adapt to the needs of information, as for most other interfaces DWH. 6. What data must be loaded into the Data Warehouse? The data from the DWH finally his place in the organization. A \ "Let \ 's load all data, security is a safe method for the design of DWH. Decisions on what should and should not be taken into consideration very early in the project easily. After the success of the good supplied, in use, for profit and DWH, there will always be somewhere on the financing of the interfaces previously ignored. Given the expected life of DWH is quite aware that some sources. The selection of data for companies must be thinking, particularly as regards the end. If it can \ 't, as the data are profitable to operate, they stay! See also Tip # 3 7th Data Warehousing and Business Policy 8th Data Warehouse projects Traps Some often repetitive "lock" on the road to a handover of a Data Warehouse: * Some data are necessary, but not for are not available, or can not * Maintenance of tuning, indexing, and backup and recovery, mostly below -- * The different possibilities for the calculation of the same phenomenon at different results, and no one is able to coherently explain the difference (s) * The lack of metadata, development and spending too much time to discover what is really a \ "is \" 9th DWH set of hardware and software In data warehousing, it is not the hardware, not software: it is the seamless integration of both. Those in the first place, or at the end of their project, paid a high price for that mistake. The reasons are as follows: ท from a project management that you never want to be among the suppliers, whether a proposed solution doesn \ 't as expected ท tuning and indexing of the database is very important and very complex, of necessity, left to specialists (in-house training) 10th Performance is the key Although don \ 't technology often factors contributing to this important in the Data Warehouse, for admission, no other factor is as important as performance. When the size increases over time, this factor is even more important. There are three reasons: 1. Performance has a great influence on the speed (the first load is increasingly a lot of time), and for the duration of delivery time on DWH 2nd Performance acceptance or break the end user, including the predictability of performance 3rd Performance has a huge influence on the productivity of end users, the last driver of pay-off