作者：
Thanos, C.Carlesi, C.Bertino, E.CNR
Istituto di Elaborazione della Informazione Pisa Italy CNR Istituto di Elaborazione della Informazione Pisa Italy
In this paper the relationship between average transaction response time and the static degree of interference for four concurrency control algorithms based on two-phase locking is established. In particular the incre...
详细信息
ISBN:
(纸本)0818605014
In this paper the relationship between average transaction response time and the static degree of interference for four concurrency control algorithms based on two-phase locking is established. In particular the increase in average response time has been measured with respect to the static degree of interference for a given multiprogramming level, database and transaction size. Also the increase in average probability of a lock request conflicting with another one has been measured with respect to the static degree of interference. In addition the average number of aborted transactions has been measureed for each algorithms.
When a data item of a database system is updated, a typical reliable storage update mechanism performs three or four secondary storage write operations in order to update the data redundantly stored on secondary stora...
详细信息
ISBN:
(纸本)0818605014
When a data item of a database system is updated, a typical reliable storage update mechanism performs three or four secondary storage write operations in order to update the data redundantly stored on secondary storage. Thus, updating the data on secondary storage may become the bottleneck for transaction processing. We show that internal processing of transactions can be separated from secondary storage update operations, and that the database state defined by the secondary storage data can be maintained consistent if updates of transactions are moved to secondary storage according to the 'U-precedence' defined for the transaction that created them. Therefore, secondary storage update operations can be buffered as long as this requirement is satisfied. Secondary storage update buffering can enhance the performance of databasesystems.since it allows more concurrency and flexible disk scheduling.
Failures in a distributed system (such as node crashes) can give rise to unwanted computations referred to as orphans. Orphans can interfere with executions of other programs, thus giving rise to unpredictable behavio...
详细信息
ISBN:
(纸本)0818605014
Failures in a distributed system (such as node crashes) can give rise to unwanted computations referred to as orphans. Orphans can interfere with executions of other programs, thus giving rise to unpredictable behavior. In this paper a graph model of computation is utilized to discuss the orphan phenomenon in a rigorous manner. In particular, conditions are derived for interference-free executions of programs. In a distributed system consisting of 'clients' and 'servers', where clients interact with servers by making use of remote procedure calls, various techniques for treating orphans under 'at least once' and 'exactly once' call semantics are next investigated.
We describe a reliability algorithm being considered for DDM, a distributeddatabase system under development at Computer Corporation of America. The algorithm is designed to tolerate clean site failures in which site...
详细信息
ISBN:
(纸本)0897910974
We describe a reliability algorithm being considered for DDM, a distributeddatabase system under development at Computer Corporation of America. The algorithm is designed to tolerate clean site failures in which sites simply stop running. The algorithm allows the system to reconfigure itself to run correctly as sites fail and recover. The algorithm solves the subproblems of atomic commit and replicated data handling in an integrated manner.
An experimental local area network (ADNET) has been developed at the Admiralty Surface Weapons Establishment (ASWE) to investigate the use of a distributed computer architecture for shipborne command and control syste...
详细信息
An experimental local area network (ADNET) has been developed at the Admiralty Surface Weapons Establishment (ASWE) to investigate the use of a distributed computer architecture for shipborne command and control systems. The ADNET Communications System provides the means by which software processes distributed throughout the network can communicate. At the hardware level it uses the ASWE Serial Highway. This is a polled multi-drop bus which employs hardware redundancy to achieve high reliability. Both broadcast and point-to-point intercomputer message transfers are supported and a high level of error detection and correction is provided. The software of the communications system provides protocols which eliminate the need for a fixed connectively within the network. This allows the applications software to be dynamically reconfigurable and leads to the provision of fault tolerant functions for the network. The ASWE Serial Highway is described together with the software and protocols of the communications system. In addition, the use of the protocols by fault tolerant functions is illustrated by examining the operation of the ASWE distributeddatabase Management System (ADDAM).
This symposiumproceedings contains 26 papers dealing with reliability problems in distributeddatabasesystems.(DDBS). These problems seem to appear often as paradoxes, as the reason for the distribution is the reduc...
详细信息
This symposiumproceedings contains 26 papers dealing with reliability problems in distributeddatabasesystems.(DDBS). These problems seem to appear often as paradoxes, as the reason for the distribution is the reduction of vulnerability associated with failures in simple centralized computers. The replication of hardware permits some parts of the system to operate while other parts have failed. The replication of software and major parts of databases is considered in order to reduce the communication overhead. Issues of error confinement, optimistic concurrency control and nested multilevel transactions, two-phase locking (2PL), read only transactions and transaction atomicity are discussed. Three papers are in extended abstract form and two in abstract form only.
作者：
Bhargava, BharatUniv of Pittsburgh
Dep of Computer Science Pittsburgh PA USA Univ of Pittsburgh Dep of Computer Science Pittsburgh PA USA
It is shown that the optimistic approach to concurrency control lends itself naturally towards a design of a reliable distributeddatabase system. In addition, the functions of concurrency control, integrity control, ...
详细信息
It is shown that the optimistic approach to concurrency control lends itself naturally towards a design of a reliable distributeddatabase system. In addition, the functions of concurrency control, integrity control, and atomicity control can be integrated. This integration can provide a higher degree of concurrency and continuity of operations.
Evidence of distribution, such as work distribution, distributed file and data base management, etc. , creates a number of reliability issues that must be addressed by the system software. Many of these are similar to...
详细信息
Evidence of distribution, such as work distribution, distributed file and data base management, etc. , creates a number of reliability issues that must be addressed by the system software. Many of these are similar to problems arising in centralized systems.however, there are a number of new issues as well as complications of common ones. Some of the specific reliability issues raised by the organizational and operational characteristics of fully distributed processing systems.(FDPS's) are outlined. It is important to bear in mind that decision making is decentralized and that there are no permanent master-slave relationships in an FDPS.
暂无评论