摘要
通信和数据库技术的发展，已经改变了用户处理数据方式。当前数据处理的特点是日渐增长的应用需求，要求存取预先存在的局部数据，这些数据分布在异构软、硬件的网络节点上。一个多数据库系统MDBS允许用户存取多个分布的、异构的、自治的数据库(DBMS)中的数据。在多数据库中，全局事务是在MDBS的控制下执行，局部事务在局部DBMS控制下独立执行。不同的局部数据库可能采用不同的事务处理机制。每个局部DBMS对其上的全局子事务和局部事务有完全的控制权，如在事务的任何执行点上终止事务。典型地，局部数据库应该没有设计和结构的变化。局部数据库之间彼此独立，互相不能协调，因而传统同构系统事务处理技术如原子提交和事务一致性不适合多数据库，我所作的工作理论上探讨多数据库事务的原子提交策略和一致性策略，编程上探讨原型系统Panorama的实现和查询处理服务器的实现。
第一，探讨多数据库事务的原子提交策略，同构系统的各种方法不适合异构系统，因为这些方法需要特殊的提交过程，而已存在的局部数据库事务管理器不支持它们。系统阐述了两种提交策略：先提交策略和后提交策略，它不改变现有的事务管理器。其中，后提交策略需要在已存在的局部数据库增加并发控制和恢复部件，增加了运行负荷。提出多级事务是为了在集中数据库中增加并发度，如果多级事务用在异构系统中，两种提交策略重用该事务模型中的并发和恢复机制，以便使全局事务原子提交。特别值得一提的是：先提交策略与多级事务的结合，还能增加事务处理的并发度而不增加运行负荷。
第二，研究了多数据库事务保持一致性的问题。全局串行化是传统数据库事务一致性准则。但是在多数据库中保持全局串行化很难。在多数据库环境下，实现全局串行化的主要困难在于解决多数据库事务之间的间接冲突，局部事务很难解决间接冲突，因为局部事务的行为甚至局部事务本身，全局事务并不知道。为了解决这个问题，在全局事务的每个子事务之间增加数据操作，以便在每个成员数据库的全局子事务之间创造直接冲突，典型的方法是ticket算法。ticket算法的条件是局部数据库只保证局部数据库的串行化执行，它不侵犯局部自治性，但能保证全局事务的串行化。它的缺点是并发度很低。在多数据库中解决低并发度的一种办法是放松串行化的要求，如两级串行化2LSR，在多数据库中确保2LSR执行是比较简单的，并能取得很高的并发度。应用该算法和多数据库的完整性约束，根据实际情况，划分不同的多数据库模型，并能保持事务的一致性。划分的模型覆盖了很多多数据库实际环境。
第三，阐述了CORBA规范。在多数据库中用CORBA作为丈撑平台，有很多优点，CORBA能够处理平台级的异构性，并能提供位置和实现的透明性，换句话说，对象的实现和对象位置的改变对客户端无任何影响，它减少了客户端代码的复杂性，不改变客户端代码，客户端能发现新类型对象，CORBA的这种结构十分有用，在注册新的DBMS并不影响已存在的系统，它也能减少开发的代码量，而且，它能提供很多在数据库级别处理异构性功能和一些在应用级别处理交互性的功能。
第四，系统描述了一个基于CORBA的多数据库原型系统Panorama实现给出了Panorama中的主要设计思想和使用的基本技术策略，如Panorama体系结构、模式集成、基本实现策略、查询处理、事务处理、客户端程序等，Panorama中集成了三个数据库ORACLE、SYABASE、DM2。目前的实现允许用户通过一种基于SQL标准的全局查询语言，存取局部数据库中的数据，到目前为止，它支持分布式计算，跨平台操作，不指明数据源，并发机制，如多用户同时进行select操作，能处理update语句，它还能处理嵌套事务。
9
”最后，介绍了多数据库原型系统Palxxallla，查询处垣u巳易器的设计和实现，给出了查询处 良
务器的所使用的数据结构，输入输出参数，查询处酬良务器的功能描述和工作流程描述；并分析了
Panon的优点和ie6．，指明了今后的努力方向。
Recent progress in communication and database technologies has changed the user riata processing environment .The present data processing situation is characterized by a growing number of applications that require access to various preexisting local data source located in heterogeneous hardware and software environments distributed among the nodes of a network .A. multidatabase (MDBS) system is a facility that allows users access to data located in multiple distributious heterogeneous autonomous database management systems (DBMS), in such a system,global transactions are executed under the control of the MDBS. Independently, local transactions are executed under the control of the local DBMS. Each local DBMS integrated by the MDBS may employ a different transaction management scheme. In addition, each local DBMS has complete control over all transactions (global and local) executing at its site, including the ability to abort at any point of the transactions executing at its she. Typically, no design or internal DBMS structure changes are allowed in order to accommodate the MDBS. Furthermore, the local DBMS may not be aware of each other, and, as a consequence, cannot coordinate their actions. Thus, traditional techniques for ensuring transaction atomicity and consistency in homogeneous distributed database systems may not be appropriate for an MDBS environment The objective of this paper is to provide a research of transactions atomicity > consistency in mulitdatabase and a mulitdalabase prototype panorama implementation on CORBAx query process server implementatioa
First we focus on atomic commitment of transactions in MDBS. The various solutions for homogeneous systems are not appropriate for heterogeneous systems, since they require specific commit procedures that are not provided by most of the existing transaction managers. We systematically discuss two alternative strategies, before commitment strategy and after commitment strategy, which do not require changes of the existing transaction managers. The first strategy requires additional concurrency control and recover)' components to be implemented on top of the existing database systems. Since these components cause additional runtime overhead, we developed a second, alternative strategy. If multi-level transactions are used for the heterogeneous system, our strategy reuses the concurrency control and recovery mechanisms of this transaction model in order to provide an atomic commitment for global transactions. Multi-level transactions have originally been developed to increase the degree of concurrency in centralized databases system; we reuse these mechanisms together with two alternative strategies to provide an atomic commitment for global transactions. Especially before commitment strategy together with multi-level transactions can increase the degree of concurrency of multidatabase and no additional overhead.
Second, the concept of serializabilhy has been the traditionally accepted consistency criterion in database systems. However, in MDBS, ensuring global serializabilhy is a difficult task. The main difficulty in enforcing global serializability in a multidatabase environment lies in resolving indirect conflicts between multidatabase
transactions. Indirect conflicts introduced by local transactions are difficult to resolve because the behavior or even the existence of local transactions is not known to multidatabase . To overcome these problems, we propose to incorporate additional data manipulation operations in the subtranstraction of each multidatabase transaction, these operations create direct conflicts between subtransaction at each participating local database system. The method is ticket algorithms, based on the approach, which requires local database system to ensure only local serializability. The method doesn't violate the autonomy of local database systems, and guarantees global serializability. its disadvantage is low concurrency. One way to overcome the problem of low concurrency is to relax the serializability requirement
引文
[1] YBreitbart,H. Garcia-Molina .A. Silberschatz. Overview of Multidatabase Transaction Management. In VLDB Journal. Vol. l,No. 2:181-293, 1992
[2] Kenneth Barker. Doctor's Paper: Transaction Management on Multidatabase Systems. URL: ww.cs.umanitoba.ca/-barker/ All_Publications.htm .1990．
[3] Batini C. et al , Comparision of methodologies for database schema integration, ACM Computer Survey, Vol. 18,No.4, 1986
[4] 郑振楣、于戈、郭敏著，分布式数据库，科学出版社，1998年6月。
[5] J.M. Smith et al.MULTIBASElntegrating Heterogeneous Distributed Database Systems,In Proc. National Computer Conference. 1981 May, pp. 487-499
[6] M.TEMPLETON,Mermaid-A Front-End to Distributed Heterogeneous Databases. Proc. IEEE May 1987,Vol75 No.5,pp:487-708
[7] N.Goodman. P. A. Berstern, E.Wong te al. Query Processing in SDD1 :A System for Distributed Databases ACM-TODS, 1981,6:4
[8] Litwin et al. SIRIUS Systems for Distributed Data Management. In Distributed Data Bases. H.-J.Schneider(ed.),Amsterdam:North-Holland.1982,pp:311-366．
[9] VGligor .R.Popescu_Zeletin. Transaction Management Systems. IEEE Computer. Jan 1984 Voll7 No.l pp :33-43．
[10] P.P.S Chen,The Entity-Relationship Model:A Basis for the Enterprise View of Data,AFIPS Conf. Proc. Vol.46, AFIPS Press,NJ,1977．
[11] VGligor, et al. Interconnecting Heterogeneous Database Management Systems. IEEE Computer Jan 1984 Vol 17No.l pp:33-43
[12] Batini C. et al , Comparision of methodologies for database schema integration, ACM Computer Survey, Vol. 18,No.4,1986
[13] C.Pu.Superdatabases:Transactions across database boundaries, Q. Null. IEEE Transaction on Data Engmeering,September 1987．
[14] C.Pu.Superdatabases for composition of herterogeneous databases. In Proc. 4th International Conference on Data Engineering. Feb. 1988．
[15] K.Eswaran ,J. Gray et al. The notion of consistency and predicate locks in a database systems. CACM, 1976Vol.19,No.1,
[16] P.A. Bernstein. V Hadzilaws , N. Goodman . Concurrency control and recovery in database systems. Addison-Wesley, 1987
[17] Y.Breitbart,P.olson and G. Thompson .Database integration in a distributed heterogeneous database system. In Proc. of the 2nd International Conference on
DataEngineering.Feb 1986.
[18]Y. Breitbart and P.Paolini.Session chairman overview-the multibase session,Distributed Data Sharing Systems. 1985.
[19]Y. Breitbart and A. Silberschatz. Multidatabase update issuesProc,of the ACMSIGMOD International Conference on the Management of Data. June 1988.
[20]Y. Breitbart,A.Silberschatz and G.Thompson.An update mechanism for multidatabase systems.Q.Null,IEEE Transaction on Data Engineering,Sep 1987.
[21]W. Du and AElmagarmid.Quasi serializability:a correctness criterion for global concurrency control in Interbase,Poc.13th International Conference on very large Databases,Aug 1989.
[22]P. chrysamhis. Synthesis of Extended Transaction Model using ACTAACM TODS,1994,Vol.19,NO.3
[23]Umeshwar D.,Meichun H.,Rivka L..A Transactional Model for Long-Running Activities.Proceedings of the 17th International Conference on VLDB
[24]Been C.,Bernstein P.A.,Goodman N. A.. A Model for Concurrency in Nested Transaction Systems. Journal of th ACM,Vol.36,No.2,1989
[25]Calton Pu. et al. Split-transactions for open-ended activities Proceedings of 14th VLDB Conference,1989
[26]Dennis S. Transaction Chopping:Algorithms and Performance Studies,ACM TODS,1995,Vol.20,No.3
[27]A.Elmargarmid,Y. Leu et al,A Multidatabase Transaction Model for Interbase.Proc. Of the 16th VLDB,Aug.1989
[28]M Rusinkiewicz,A.Elmagarmid and Y. Leu et al.Extending the transaction model to capture more meaning.SIGMOD record,Vol, 19,No.1.March 1990
[29]周源源等，可扩充事务模型．计算机科学。1993．Vol．20．No．3
[30]金远平。王能斌。联邦数据库管理系统FDBMS的设计与实现．计算机学报．1993 Vol 16No．6．
[31]王宁，陈滢，俞本权，徐宏炳，王能斌．一个基于CORBA的异构数据源集成系统的设计．软件学报．1998 Nay，Vol 9，No．5
[32]于戈，王国仁，郑怀远．CIMS多数据库中的弹性事务管理方法第十一届全国数据库会议论文集，1993．9．
[33]石祥滨，张斌，于戈，杨晓春，靳莉莉，郑怀远．基于CORBA的异构分布信息系统．小型微型计算机系统．1997 Feb．Vol 18：No．2
[34]石祥滨，张斌，王国仁，于戈，郑怀远．一个实现对象查询语言的形式化基础．软件学报．Vol．9，No．5 Nay 1998
[35]石祥滨，张斌，于戈，郑怀远．基于模式集成语义的查询处理．软件学报1998 May Vol．9，No．5
[36]姚卿达，张耕，邓豫．联邦数据库GD-MDBS的总体设计与模式集成，软件技术．
[37]肖卫军．肖卫军硕士毕业论文．华中理工大学1998
[38]Sena Nural,Pinar Koksal,Fatma Ozcan,Asuman Dogac.Query Decomposition and Processing in Multidatabase Systems. In proc.Of Workshop on Nest Generation Information Technologies and Systems,Naharia,Israel, 1995b.
[39]A.Dogac,C.Dengi,E.kilic,G.Ozhan, F.Ozcan,S.Nural etc.METU Interoperable Database System,1995b,ACM Sigmoct Record,24(3).
[40]Cem Evrendilek,Asuman Dogac Query Decomposition,Optimitation and Processing in Multidatabase Systems.1995b,In proc.Of Workshop on Nest Generation Information Technologies and Systems,Naharia,Israel.
[41]A.Dogac,C.Dengi,E.kilic,G.Ozhan,F.Ozcan,S.Nural etc.A Multidatabase Systems Implementation on CORBA.1996,In 6th Intl. Workshop on Research Issues in Data Engineering(RIDE-NDS'96),New Orleans.
[42]Ugur Halici.Budak Arpinar, Asuman Dogac.Serializability of Nested Transactions in Multidatabases.http:www.srdc.edu.tr
[43]V. Gligor and R.Popescu-Zeletin.Transaction management in distributed heterogeneous database management systems.Information System,1986,Vol.11,No.4
[44](德)Th．黑德尔著，漆永新等译．数据库系统实现方法．科学出版社，1986年3月．
[45]G.Weikum.Principles and Realization Strategies of Multilevel Transaction Management.ACM Transaction s on Database Systems,Vol 16,No.1,1991
[46]D.Georgakopolous,M.Rusinkiewioz and Asheth.On Seriatizability of Multidatabase Transactions through forced local Conflicts.In Proceedings of the Seventh International Conference on Data Engineering, Kobe,Janpan
[47]D.Georgakopoulos,M Rusinkiewicz,and A,P.Sheth. Using Tickets to Enforce the Serializability of Multidatabase Transactions.IEEE Transactions on Knowledge and Data Engineering, 1994,Vol 6 No.1
[48]K.L.Wu,P.Yu,C.Pu.Divergence Control for Epsilon-Serializability. Proeeding of 8th International Conference on Data Engineering. Phoenix,AZ,February,1992.
[49]M. Ouzzani, M.A Atroun and N. L. Belkhodja. A Top-Down Approach for two level serializability.VLDB'94
[50]Object Maanagement Group:"The Common Object Request Broker:Architecture and Specification".OMG Document Number 91.12.1 ,December 1991.
[51]CORBA Services:Common Object services Specification.OMG, March 1995.
[52]The Common Object Request Broker:Architecture and Specification Revision 2.0,OMG July 1995.
[53]VISIGENIC Programmer's Guide Version 3.0: VisiBroker for C++,Visigenic Software Inc.,1997.
[54]T. Harder. KRothermel.Concurrency Control Issues in Nested Transactions. VLDB Journal,2(1), 1993.
[55]Y. Breitbart, A Silberschatz, G.R. Thompson. Reliable transaction management in a failure-prone multidatabase environment.VLDB Journal, 1(1),1992.