Four key points of the hottest Bi product selectio

  • Detail

Four key points of Bi product selection

once enterprises have a lot of data, they want to do some analysis. A few years ago, data warehouse projects were first widely carried out in the telecommunications and financial industries; In the past two years, the news of building data warehouses has also been coming from the retail and manufacturing industries. This is because these enterprises have deposited enough data

of course, some enterprises are out of the need of business analysis, while others may watch their competitors go to a data warehouse for fear of falling behind and catching up. In any case, when enterprises start their first data warehouse (or business intelligence) project, they will be troubled by a problem - what kind of product portfolio should be used

these mechanical properties have opened up a series of new uses of "foreign gadgets"

before model selection, you might as well count the existing mainstream BI products

in terms of databases, there are DB2, Oracle, SQL server, Teradata, and previously redbrick, which was specially used for data warehouse (after being incorporated by IBM, it withdrew from the historical stage)

etl tools, such as Datastage and powercenter, are relatively mainstream. In addition, many companies also have their own ETL products, such as SAS ETL server, Bo data integrator, etc

olap tools can also be subdivided into MOLAP (muiltdimension OLAP) and ROLAP (relational OLAP). The former can choose products from Hyperion, Cognos and Microsoft, while the latter has little choice. For example, MicroStrategy may have a relatively large market share at present, and it used to work with redbrick. In addition, there is a ROLAP tool called metacube, which also withdrew from the market as early as 2000

data mining products are particularly suitable for protecting outdoor steel structures in the petrochemical industry; The other is in the field of thick coating. There are two major manufacturers such as SAS and SPSS, and IBM and Teradata also have their own mining tools. In addition, there are many choices in report server and front-end tools, among which Cognos, Bo and Brio are more mainstream

most of these mainstream products are imported. There are also research and development of BI products in China, but most of them are limited to ETL, front-end and data mining products. This situation has a lot to do with the previous domestic product selection, because many projects in these three areas have been independently developed from scratch

speaking is better than singing

from the current product selection process of various users, the problems are more in the following aspects

first, we only see trees but not forests. We only want to reduce the cost of a single tool, but ignore the overall cost. Some large enterprises spend a lot of money on it construction, never caring about how much they invest in purchasing products, and each piece is the best. However, more enterprises have to consider the cost and save if they can. In addition to databases and OLAP tools, one of the things you often think about is whether it will save money if ETL and front-end are developed by showing yourself

however, the problem is that if we only consider the cost of a single product and ignore the comprehensive project cost, we are likely to be troubled by the hard to maintain procedures in the end, or even completely overturn and start over, and the cost may be higher. For those large enterprises with rich and powerful resources, even if they choose the best products in each field, the combination is not necessarily the best

second, comprehensive solutions are not popular. Nowadays, almost every company claims to be able to provide comprehensive solutions and one-stop services. IBM, Oracle and SAS all claim to be such full solution providers, that is to say, their product lines have included a large number of human factors, such as databases, OLAP, ETL and other tools. If you choose them directly, you won't have to worry about tool selection anymore. In fact, such manufacturers are gradually increasing. Some rely on OEM professional products, while others directly purchase and merge professional tool manufacturers

it may be wise to choose such a product portfolio, but it is not completely affordable in China. Because Chinese people pay attention to checks and balances, all the cakes are given to your family. In the future, if you get up, won't I be very passive? Therefore, it is usually necessary to introduce multiple original manufacturers and let them cooperate and compete

third, the evaluation report is difficult to be objective. The person in charge of product selection in an enterprise usually requires the manufacturer to provide an evaluation report, listing several options, their advantages and disadvantages, and finally arriving at which option is most suitable for him

if you literally understand this report, which aspect of a product is better and which aspect is not, this is pure nonsense. Such a report cannot be expected to be objective. Think about it. For manufacturers, this is a list. Of course, their products will go up to heaven; For integrators, perhaps they are acting for a certain product, and of course they will recommend specific products; For this person in charge of model selection, maybe he has long been interested in a certain tool, or has a good relationship with a manufacturer, or even has further transactions. In such a situation, it is not easy to be objective

I once encountered a customer's system upgrade. The original system has been used for several years. The customer hopes to change the tools together. After putting forward the plan (of course, the plan we are familiar with), the person in charge of the other party also puts forward a small request to see whether ETL tools can use xx products. Unfortunately, no one in our company knows that tool at all, and there are few application cases in China, so we dare not agree. Unexpectedly, after the meeting, the salesperson said, "you're all stupid. Isn't it obvious that people have already done their work? If we can use that product, we can use it. There's no need to offend customers."

four key points of model selection

in fact, if you choose these mainstream products, you know that "there is no best, only the most suitable". What is the most appropriate? Put aside the factors mentioned above, there are actually four points to consider - product cost, developers' familiarity with the product, whether there are similar cases, and whether you have a good relationship with other manufacturers

first look at the cost. NCR, IBM and Oracle have complete product lines, but they are expensive. Microsoft's products are cheaper, but if you have enough data, I'm afraid you don't dare to use it, let alone those open-source products that don't cost money. Of course, the cost is not only determined by the price of the product itself, but also considered as the cost of personnel learning, project delay and low customer satisfaction. These implicit costs are difficult to calculate

let's look at the experience of personnel. People's learning curve is inevitable. Don't imagine that people can become masters immediately after contacting a new product, and can make a good framework based on unfamiliar products. In this regard, Oracle and Microsoft obviously have advantages, because there are many people with experience in these two products, so it is easy to find. Of course, if your original business system uses one of these products, you might as well still use it

cases are more important than white papers. When you decide not to use one product and choose different product combinations in order to balance the interests, you should consider whether the thickness of the sample of their plastic simply supported beam impact tester is 80 10mm or not. But if you look for this kind of information from the white paper of the product, you will find that it is beautiful to say, and how they will be perfectly "seamless" compatible with each other, but in fact it is not the case. Therefore, do not look at these words, but look for similar cases in the same industry. If not, look for other industry cases with similar data volume, business complexity and similar applications

finally, it depends on the relationship between you and the manufacturer. People can't help but support you during the implementation of the project. If you are a large enterprise, fortunately, manufacturers are running behind you. If you are a small unit and the project budget is small, sorry, the sales of those large manufacturers are not willing to lower their noble heads. Therefore, it is better to observe which manufacturer may give greater support

if you still haven't decided which combination to choose from these four aspects, you may have only one last move: dice. (end)

Copyright © 2011 JIN SHI