Please use this identifier to cite or link to this item: http://hdl.handle.net/11455/18981
標題: 一個新的物件導向資料庫版本控制模型
A New Version Model for Object-oriented Databases
作者: 馮懷碧 
Feng, Hwai-Bih 
關鍵字: object-oriented database;物件導向資料庫;version model;large object;data retrieval;version difference;版本模型;大型物件;資料擷取;差異資料
出版社: 資訊科學學系
摘要: 
提供物件版本是物件導向資料庫系統不可或缺的一項重要功能.本論
文提出一個新的物件版本模型,此版本模型是針對目前既有各種版本模型
的缺點加以修改,並提供較多的運算(operation),讓使用者可以很方便的
使用物件版本功能.此版本模型對大型物件(large object)提出一個新的
表示方法來避免造成浪費儲存空間,同時也可以有效的處理一般物件(
regular object),大型物件(large object)及複合物件(composite
object)的版本. 我們已將此物件版本模型實作於CTSS(Chao Ta
Storage System),CTSS是由交通大學資料庫實驗室所發展出來的一套物件
儲存器,在此可驗證我們版本模型實作之可行性.在實作版本模型時,我們
利用模組化設計,只需對原有的系統做部份介面修改就可以將我們的版本
模型加入,使得我們的版本模型與原有的系統有較高的獨立性. 為了
瞭解我們的版本模型與其他版本模型的優缺點,我們在論文中描述了各系
統版本模型的特性,然後再將這些系統的版本模型與我們提出的版本模型
做比較.由於我們的大型物件版本以差異資料(version difference)的方
式來儲存,可節省許多儲存空間.針對各種新舊版本修改比率所節省的儲存
空間做分析,我們發現,若修改的資料大小為原始物件版本的30%,且修改資
料區段數為10時,所節省的空間至少為69.7%.而在擷取物件版本的時間方
面,當利用物件辨識碼(OID)來擷取某一個一般物件(regular object)或整
體儲存大型物件(evaluated object)時,其所需的時間與擷取不可版本性
物件所需的時間完全相同.

Providing the functionality of object version is essential to
object-orienteddatabase systems. In this study we propose a new
object version model, whichenhances the existing version models
and provides more operations for users to use the functionality
conveniently. This version model uses a new representation
scheme to avoid the waste of storage space for large objects.At
the same time, it can process regular objects,large objects and
compositeobjects efficiently. We have implemented the proposed
version model in CTSS(Chao Ta Storage System), which is a
storage system developed by the database laboratory of National
Chao Tung University. From this implementation, we have
verifiedthe feasibility of this version model. The model is
designed as an independent module, which can be easily plugged
into CTSS or other object storage systems with a minor
modification of some of its module interfaces. To understand
the advantages and disadvantages of our model over others, we
review the characteristics of various version models and then
compare these models with ours. Since large objects in our model
are stored in version difference, a fair amount of space saving
can be expected. We analyze the space saving for each different
data update ratio, and observe that, if the updatedata size is
about 30% of the original object version and there are 10
regionsupdated, then there is at least69.7% of space saving.On
the other hand, when we retrieve a regular or evaluated object
version by its OID, the retrieving time is identical that needed
for a non-versionable object.
URI: http://hdl.handle.net/11455/18981
Appears in Collections:資訊科學與工程學系所

Show full item record
 

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.