⭐ 欢迎来到虫虫下载站! | 📦 资源下载 📁 资源专辑 ℹ️ 关于我们
⭐ 虫虫下载站

📄 6.txt

📁 This complete matlab for neural network
💻 TXT
字号:
发信人: Axiao (阿肖期待涅磐中), 信区: DataMining
标  题: 再转一篇,出处同上
发信站: 南京大学小百合站 (Sat Mar 16 21:38:48 2002)

 数据仓库开发方法与传统应用的不同 by inmon  

SDLC Versus CLDS 




by W.H. Inmon 




The classical system development life cycle is one that has been described as 
a waterfall. One activity feeds into another until the end results finally flo
w out the end. The classical steps of the sdlc are: 




requirements gathering, 




analysis, 




design, 




programming, 




testing, and 




implementation. 


The steps occur in a prescribed and predictable manner. 




The end user that the world of the sdlc applies to is someone that can be call
ed a farmer. A farmer is someone who knows what his/her requirements are befor
e they set out to do an analysis. The farmer knows when a query is formulated 
what to expect. As a consequence, the sdlc fits the needs of the farmer quite 
well. 




But the sdlc does not fit the world of data warehousing well at all. In order 
to understand why, consider the data warehouse user, someone who can be called
 an explorer. The explorer is an individual who has the attitude of give me wh
at I say I want, then I can tell you what I really want. The explorer operates
 in a mode of discovery. The explorer says: "ah, now that I see what the possi
bilities are, now I can tell you what the requirements are. But until I see wh
at can be done, there is no way to tell you what I really want." 




Trying to coerce requirements out of an explorer is an exercise in futility. 





How is it that systems are built for the explorer? Systems are built in the fo
llowing manner: 




implemented data is gathered, 




the data that has been gathered is probed and tested, 




next programs are written against the data to see what is required to access a
nd analyze the data, 




after the results of the programs are collected, they are analyzed, and 




requirements are formulated. 


These steps are done iteratively, where each step may or may not feed back to 
the previous step. 




The development methodology that has been described is the same as the sdlc ex
cept for one aspect - it is backwards. The development life cycle for the expl
orer then is sometimes called the clds, or the reverse of the sdlc. 




The farmer then fits nicely with the classical sdlc, while the explorer fits w
ith the clds. Trying to force fit the wrong methodology or development approac
h with the wrong end user only results in frustration. 











--
歌 兴 其 其 以 聊 纵 乐 失 得 俱 百 唯 日 所 有 何 情 尘 人 岁 日   

以 之 理 言 遗 作 酒 之 之 之 化 年 利 落 谓 酒 须 仇 事 生 月 出    自

自 所 颇 虽 世 此 狂 最 何 何 尘 之 是 日 消 且 执 难 颇 苦 磋 日    遣

遣,至,重,微,俗,篇,呼,者,忧,喜,土,后,图,出,磨,饮,着,了,多,短,跎,落

欢迎造访http://xiaoysh.home.chinaren.com (新地址)  有更多我的作品



※ 来源:.南京大学小百合站 http://bbs.nju.edu.cn [FROM: 172.16.5.62]

⌨️ 快捷键说明

复制代码 Ctrl + C
搜索代码 Ctrl + F
全屏模式 F11
切换主题 Ctrl + Shift + D
显示快捷键 ?
增大字号 Ctrl + =
减小字号 Ctrl + -