Please use this identifier to cite or link to this item: http://hdl.handle.net/10603/16640
Title: Synchronization based on data flow graphs for distributed and replicated databases
Researcher: Krishna Reddy, Polepalli
Guide(s): Bharadwaj, K K
Bhalla, Subhash
Keywords: Computer Science
graphs
Replicated
databses
flow graphs
Upload Date: 3-Mar-2014
University: Jawaharlal Nehru University
Completed Date: 1993
Abstract: In this study, we present concurrency control algorithms based on data flow graphs for newlinedistributed and replicated databases. newlineGraphical representations such as token models and structure models are widely applied newlineto represent data flow programs. Data flow programs are represented by directed graphs called newlinedata flow graphs where nodes represent the operations to be performed and arcs represent newlinethe data dependencies between them. The scheduling of operations is constrained by the data newlinedependencies identified by the graph. So, data flow scheduling mechanism could execute newlineoperations based upon the dependency constraints defined by the data flow graph. In this newlineway, concurrency is achieved through graphs. newlineIn this study we explore t~~ use of data flow graphs to design concurrency control newlinealgorithms. Precedence graphs are often used to determine serializability, and wait-for-graphs newlineare often used to detect deadlocks. Among these, acyclic precedence graphs indicate newlineserializability, and acyclic wait-for-graphs imply freedom from deadlocks. The tools for newlineserializability theory, and transaction management coupled with data flow graphs will provide newlinesolutions that are more efficient than conventional concurrency control techniques, such as, newlinetwo-phase locking. In the same light, we apply data flow graphs for concurrency control for newlinedistributed, and replicated databases. newlineAt first, we propose a concurrency control algorithm based on data flow graphs for a newlinedistributed database system. The concurrency control approaches based on locking, require newlineadditional efforts in deadlock detection and its elimination. The deadlock resolution protocols newlineassociated with deadlock detection algorithms abort (restart) some transactions in the deadlock newlinecycle. The restarted transaction must release all its locks and send out the requests again. The newlinepossibility of a deadlock is also connected to existence of unpredictable delays and repeated newlinerestarts of transactions in a deadlock cycle.
Pagination: iv, 164p.
URI: http://hdl.handle.net/10603/16640
Appears in Departments:School of Computer and System Science

Files in This Item:
File Description SizeFormat 
01_title.pdfAttached File125.39 kBAdobe PDFView/Open
02_certiificate.pdf158.92 kBAdobe PDFView/Open
03_abstract.pdf166.99 kBAdobe PDFView/Open
04_acknowledgements.pdf134.38 kBAdobe PDFView/Open
05_contents.pdf174.18 kBAdobe PDFView/Open
06_list of abbrevations.pdf158.93 kBAdobe PDFView/Open
07_list of graphs.pdf176.86 kBAdobe PDFView/Open
08_chapter 1.pdf687.66 kBAdobe PDFView/Open
09_chapter 2.pdf651.29 kBAdobe PDFView/Open
10_chapter 3.pdf933.31 kBAdobe PDFView/Open
11_chapter 4.pdf1.06 MBAdobe PDFView/Open
12_chapter 5.pdf793.44 kBAdobe PDFView/Open
13_chapter 6.pdf262.02 kBAdobe PDFView/Open
14_appendix.pdf881.16 kBAdobe PDFView/Open
15_bibilography.pdf451.62 kBAdobe PDFView/Open


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