適用於企業計算之軟體工廠模式

THUIR > College of Engineering > Department of Computer Science and Information > National Research Council Research Report > Item 310901/10192

Please use this identifier to cite or link to this item: http://140.128.103.80:8080/handle/310901/10192

Title:	適用於企業計算之軟體工廠模式
Other Titles:	Software Factory Model for Enterprise Computing
Authors:	周忠信,鄭有進 Jwo, Jung-Sing；Cheng, Yu-Chin
Contributors:	行政院國家科學委員會東海大學資訊工程與科學系
Keywords:	軟体工程；軟體工廠；生產工具；生產管理；達交預估PDF created software engineering； software factory； production tools； production management；fulfillment prediction and estimation
Date:	2008
Issue Date:	2011-06-23T05:57:25Z (UTC)
Abstract:	在軟體工程領域中，將軟體開發視為工廠生產模式的「軟體工廠」，早在70年代就被提出。然而，軟體工廠的概念並未真正獲得認同，究其原因則在於軟體的應用領域既廣泛而複雜，而工廠生產機制通常卻僅能生產一定類型的產品。隨著企業計算越來越重要、也越來越普及，企業應用軟體的需求越來越多，而能夠等待的時間卻越來越短。一個能夠讓企業軟體開發變得更快速、同時品質能夠更好的機制，是企業計算領域中的關鍵需求。雖然軟體工廠有其侷限，但是由於企業計算中的軟體，在架構上、操作上、與應用上，皆有一定程度的類似性。因此，若能將工廠同類型產品的大量生產機制以及工廠管理思維，應用在企業計算領域上，將可有效改善企業目前的e化成本與時程。本計劃根據企業計算所需的軟體需求，首先定義企業軟體的基本原物料組成概念–SoftBOM。同時根據基本原物料組成以及產品模型與組裝概念，發展軟體工廠的生產線架構，並進而提出一個完整的軟體工廠模式。在此工廠模式中，除了生產相關的工具外，與生產管理相關的系統，例如產品設計資料管理系統、工廠執行系統、生產規劃系統、派工管理系統、顧客服務系統等，更是落實軟體工廠正常運作的必要組成。其中，為了確保軟體工廠可以準確規劃與派工，達交預估系統更是扮演重要角色。軟體成本預估領域已有數學模型或其他傑出的研究，然而從工廠角度來看，存在人、工具、生產經驗與生產法則等關聯，傳統的硬體工廠在達交預估上多為一個智慧型的規則導向系統。本計劃在軟體工廠模式發展完成後，將根據硬體工廠的達交預估機制，實際發展符合軟體工廠模式的達交預估模擬系統。當軟體工廠的生產機制、管理系統、以及達交預估完成後，本計劃將可在企業計算領域上，開創一個有效的企業軟體大量生產方式。 As early as in the 1970’s, the notion of software factory that treats software development as a kind of factory production had been proposed. However, software factory has yet to win recognition and acceptance by the software development community. The causes of failure are mainly attributed to the widespread use of software in all sorts of disparate application domains and the ever increasing complexity of software therein. The two causes prevent the factory production model, which has otherwise proven highly effective in producing fix-type products, for software development. On the other hand, enterprise computing is becoming pervasive and indispensable. The requirements are ever increasing with the time-to-market ruthlessly compressed. Therefore, it is critical for enterprises to adopt a mechanism that enables rapid development of quality software. In researching for such a mechanism, we observe that the software factory can become a highly viable model for developing complex enterprise computing software in short time frames. The viability of software factory is founded by the observations that most enterprise computing applications share high similarities in architecture, operation, and use, which can be leveraged to eliminate the two limiting causes. The high similarities enable a software mass-production process that is amenable to effective factory management. In the project, enterprise computing requirements are captured as software bills of materials, or SoftBOMs. Based on the notion of assembly of SoftBOMs into products, an architecture is developed for software factory production that contributes to a comprehensive software factory. In addition to the various production tools used by workers in the software factory, an assortment of production management systems for product design, factory execution, production planning, job dispatching and management, and customer relation management are specified for the smooth operation of software factory. Among them, the fulfillment prediction and estimation is pivotal to accurate planning and dispatch. To this end, we will adopt the rule-based systems used by computer hardware manufacturers instead of existing mathematical models. Most notably, the rule-based fulfillment prediction and estimation captures associations among workers, tools, production rules and production experiences. We envision that, once fully developed, the production, management, and fulfillment prediction and estimation of the proposed software factory will create a new effective way for mass-producing enterprise computing applications.
Relation:	研究編號：NSC97-2221-E029-010 研究期間：2008-08~ 2009-07
Appears in Collections:	[Department of Computer Science and Information ] National Research Council Research Report

Files in This Item:

There are no files associated with this item.

Loading...