O. Ulusoy. A Study of Two Transaction Processing Architectures for Distributed Real-time Database Systems. Journal of Systems and Software, 31(2):97-108 (1995).  Home/Search   Document Details and Download   Summary   Related Articles   Check

....time burden on the performance of a MDRTDBS and it also can seriously affect the performance of the adopted concurrency control protocol. Although the concurrency control protocols proposed for DRTDBS can be extended for MDRTDBS, their performance may be very different from that in a DRTDBS [31], due to the unique characteristics of mobile network. Compared to wired networks, mobile networks are much slow, unreliable, and unpredictable. The mobility of clients affects the distribution of workload in the network and the system. Disconnection between mobile clients and base stations is ....

....between a mobile client and its base station every time when a channel request is made. Transient errors in communication are modeled by a noise factor, which affects the strength of radio signal received by the base stations and mobile clients. In the system level, a DRTDBS model is implemented [31] in which the Distributed High Priority Two Phase Locking (DHP 2PL) is used for concurrency control. The database system at each base station is shown in Figure 5. It consists of a scheduler, a CPU, a ready queue, a local database, a lock table, and a block queue. It is assumed that the database ....

O. Ulusoy. A Study of Two Transaction Processing Architectures for Distributed Real-time Database Systems. Journal of Systems and Software, 31(2):97-108 (1995).

....unless specified otherwise. 3. 2 Distributed RTDB Systems RTDB systems find applications in many areas like aerospace and military systems, computer integrated manufacturing, robotics, nuclear power plants, traffic control systems, stock market, telephone switching systems, and network management [Son90, Ulu94a]. Many of these applications, especially in the areas of stock market, communication systems and military systems, are inherently distributed in nature. Unfortunately, till now, the major focus of the research in RTDB technology has been on centralized systems. Incorporating distributed data into ....

Ozgur Ulusoy. A Study of Two Transaction Processing Architectures for Distributed Real-Time Database Systems. Technical Report BU-CEIS-94-22, Department of Computer Engineering and Information Science, Bilkent University, 1994.

....processing, usually in the form of transaction completion deadlines. Such applications include aerospace and military systems, computer integrated manufacturing, robotics, nuclear power plants, traffic control systems, stock markets, telephone switching systems, and network management [Son90, Ulu94a] The performance goal in these systems is to maximize the number of transactions that complete before their deadlines expire. That is, the important issue is whether or not a transaction completes before its deadline, but not how soon it completes before its deadline. Therefore, in contrast to ....

....deadline RTDB system, unless specified otherwise. RTDB systems find applications in many areas like aerospace and military systems, computer integrated manufacturing, robotics, nuclear power plants, traffic control systems, stock market, telephone switching systems, and network management [Son90, Ulu94a] Many of these applications, especially in the areas of stock market, communication systems and military systems, are inherently distributed in nature. Unfortunately, till now, the major focus of the research in RTDB technology has been on centralized systems. Incorporating distributed data into ....

Ozgur Ulusoy. A Study of Two Transaction Processing Architectures for Distributed Real-Time Database Systems. Technical Report BU-CEIS-9422, Department of Computer Engineering and Information Science, Bilkent University, 1994.

....a long time due to a long delay in completing the commit procedure. As an example, let us look at the twophase commit protocol (2PC) 9] which is one of the most commonly used commit protocols for traditional distributed database systems [3, 9] and was also widely used in the studies of DRTDBSs [1, 5, 6, 10, 11, 12]. In the 2PC, the processes of a transaction at different sites are divided into two groups. One of the processes is the coordinator and the others are the participants [3] The following factors can cause a long delay in the execution of the 2PC: Uneven distribution of transactions over the ....

....will be allowed to borrow the value from the committing transaction. In [2] it was found that the best range for the healthy factor is between 1 to 2 and its performance with this range is similar. 5.1.1. Transaction deadline assignment Following the assumptions described in previous works [5, 6, 8, 12, 13, 15, 20, 24], the transaction deadlines are assumed to be proportional to their expected execution 684 K. Y. LAM et al. and are defined as follows: Deadline T ar (T stage1 T stage2 ) slack factor. The slack factor (SF) is a random variable uniformly distributed between two bounds. T ar is the ....

Ulusoy, O. (1995) A study of two transaction processing architectures for distributed real-time database systems. J. Syst. Softw., 31, 97--108.

....long time due to a long delay in completing the commit procedure. As an example, let us look at the two phase commit protocol (2PC) 9] which is one of the most commonly used commit protocols for traditional distributed database systems [3, 9] and also it was widely used in the studies of DRTDBS [1, 5, 6, 10, 11, 12]. In the 2PC, the processes of a transaction at different sites are divided into two groups. One of the processes is the coordinator and the others are the participants [3] The following factors can cause long delay in the execution of the 2PC: Uneven distribution of transactions over the ....

....borrow the value from the committing transaction. In [2] it had found that the best range for the healthy factor is between 1 to 2 and its performance with this range is similar. 5.1. 1 Transaction Deadline Assignment Local Database 22 Following the assumptions described in the previous work [5, 6, 8, 12, 13, 15, 20, 24], the transaction deadlines are assumed to be proportional to their expected execution time 1 and are defined as follows: Deadline = T ar (T stage1 T stage2 ) slack factor The slack factor (SF) is a random variable uniformly distributed between two bounds. T ar is the arrival time of ....

Ulusoy, O. (1995) A Study of Two Transaction Processing Architectures for Distributed Real-time Database Systems. Journal of Systems and Software, Volume 31, Number 2, pp. 97-108.

....the transaction shipping approach, a detailed simulation program of a MDRTDBS is implemented according to the MDRTDBS model defined in Section 2. 5. 1 Model Parameters and Performance Measures The deadline of a transaction, T, is defined according to the expected execution time of a transaction [1,2,3,8,9] such as: Deadline = ar(T) pex(T) 1 SF) where SF : the slack factor is a random variable uniformly chosen from a slack range; ar(T) the arrival time of transaction T; pex(T) the predicted execution time of T. It is defined as: pex(T) T lock T process T update ) N oper where N ....

O. Ulusoy, " A Study of Two Transaction Processing Architectures for Distributed Real-time Database Systems", Journal of Systems and Software, vol. 31, no. 2, pp. 97-108, 1995.

....the performance of the transaction scheduling algorithms. Concurrency Control Techniques with Serializability: Concurrency control techniques for real time databases that use serializability as the correctness criteria include lock based protocols such as two phase locking and its variants [3, 1, 7, 60, 86, 88, 179, 180, 193, 214, 216], optimistic concurrency control protocols [74, 78, 88, 121] and timestamp ordering protocols [133, 195, 215] Using any of these techniques, conflicts between two real time transactions or between one real time transaction and a set of real time transactions are detected. For lock based ....

Ulusoy, O., "A Study of Two Transaction Processing Architectures for Distributed RealTime Database Systems," The Journal of Systems and Software (to appear), 1995.

.... et al. 1995] Chen Gruenwald 1994] Fortier et al. 1994] Graham 1993] Hou et al. 1989] Huang et al. 1989] Lin 1989] O Neil Ramamritham 1992] O Neil Ramamritham 1995] Ozsoyo glu et al. 1990] Ozsoyo glu et al. 1992] Soparkar et al. 1992b] Stankovic Zhao 1988] Ulusoy 1995b] 2.4 Disk Scheduling Similar to traditional database systems, an important candidate for performance improvement in realtime database systems is the I O subsystem. The disk scheduler in a real time database system primarily concerns the timing constraints of transactions in processing their ....

O.Ulusoy `A Study of Two Transaction Processing Architectures for Distributed RealTime Database Systems', to appear in Journal of Systems and Software, vol.31, no.2, 1995, pp.97-108.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC