Peter A. Franaszek, John T. Robinson, and Alexander Thomasian. Concurrency control for high contention environments. ACM TODS, 17:304--345, June 1992.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....could be leveraged to improve performance [23] Unlike my approach, almost all of these approaches, as well as the four approaches discussed above, require special hardware support. The exception, and the prior work closest in spirit to my own, is a proposal by Franaszek, Robinson and Thomasian [12]. In addition to requiring no special hardware support, their proposed approach also uses speculative execution to generate I O prefetches. More specifically, they proposed that databases be implemented such that, whenever a transaction is unable to execute normally because some necessary lock is ....

Peter A. Franaszek, John T. Robinson, and Alexander Thomasian. Concurrency control for high contention environments. ACM Transactions on Database Systems (TODS), 17(2):304--345, June 1992.

....is very small in a database system that is not thrashing. Performance There are many papers available that discuss the performance of the various concurrency control protocols. In particular, two phase locking is prone to thrashing and this phenomenon is discussed and analyzed in [TGS85, FRT92, FR85, Tho93] A load control scheme based on a half and half rule to avoid thrashing is proposed and evaluated in [CKL90] Half and half rule is based on the hypothesis that the system starts thrashing when more than 50 percent of the transactions in the system are waiting for the data locks ....

Peter A. Franaszek, John T. Robinson, and Alexander Thomasian. Concurrency Control for High Contention Environments. ACM Transactions on Database Systems, 17(2), 1992.

....Consequently, it has been suggested to enhance two phase locking by means of transaction aborts when a specified limit for the maximum wait depth is exceeded. This idea has been elaborated in two variants that are known under the names cautious waiting [4] 44] and wait depth limitation [30] [31]. Finally, one can also conceive using global metrics like throughput and response time as indicators of data contention. A drop in throughput or an increase of response time would be interpreted as an increasing degree of data contention. In fact, it has been proposed to exert load control on ....

.... the conflict driven load control method that exploits the same knowledge as the INF method but differs from the INF method in that it serves the system entry queue in a non FIFO manner whenever this is estimated to be beneficial, WDL: the wait depth limitation method proposed by Franaszek et al. [31] where the depth of a transaction waiting chain is limited to 1, HALF: the half and half method proposed by Carey et al. 10] MPL: a global limitation of the MPL where the limit is tuned manually by exhaustive trials for specific experiments, NO: a strawman where no load control is exerted ....

P. Franaszek 7 J. T. Robinson and A. Thomaslan. Concurrency Control for High Contention Environments. ACM Transactions on Database Systems 7 17 (2)7 304-345 (1992).

....execution of conflicting transactions. Various 2PL lock conflict resolution methods have been proposed in the literature in the past, including those that can cause deadlocks, e.g. general waiting [2] and those that are deadlock free (due to selective transaction aborts) e.g. running priority [4, 5], cautious waiting [6] and wait depth limited methods [5, 7] If the system allows deadlocks to exist, it requires the use of a deadlock detection algorithm. One common strategy is to check the wait forgraph (WFG) whenever a transaction is blocked. If the blocked transaction is involved in a ....

.... resolution methods have been proposed in the literature in the past, including those that can cause deadlocks, e.g. general waiting [2] and those that are deadlock free (due to selective transaction aborts) e.g. running priority [4, 5] cautious waiting [6] and wait depth limited methods [5, 7]. If the system allows deadlocks to exist, it requires the use of a deadlock detection algorithm. One common strategy is to check the wait forgraph (WFG) whenever a transaction is blocked. If the blocked transaction is involved in a deadlock cycle, then one transaction in the cycle is aborted to ....

Franaszek, P. A., Robinson, J. T. and Thomasian, A. (1992) `Concurrency control for high contention environment.' ACM Trans. Database Syst., 17, 304--345.

....4 presents the results and analysis of a detailed performance study on the Charm prototype. Section 5 concludes this paper with a summary of main results from this research and an outline of on going work. 2 Related Work The idea of TSTE was originally proposed in [14] and has been analyzed in [12, 13], which exploited the concept of access invariance to perform speculative disk I O, and compared TSTE with various combinations of optimistic concurrency control and two phase locking through a simulation based study. None of these works included optimistic TSTE and hybrid TSTE schemes discussed ....

Franaszek, P.A. et al. "Concurrency control for high contention environments," ACM Transactions on Database Systems, June 1992. vol.17, no.2, p. 304-45.

.... algorithms in firm deadline RTDBS [HCL92] firm deadlines means that transactions which miss their deadlines are considered to be worthless and are immediately discarded from the system without being executed to completion) In contrast, locking protocols such as 2PL [Eswa76] and WDL [FRT92] provide considerably more throughput than optimistic algorithms for standard transactions in conventional resource limited DBMS [ACL87, FRT92] Therefore, in general, standard transactions may expect to receive poor service from real time database systems that have been designed with only ....

.... to be worthless and are immediately discarded from the system without being executed to completion) In contrast, locking protocols such as 2PL [Eswa76] and WDL [FRT92] provide considerably more throughput than optimistic algorithms for standard transactions in conventional resource limited DBMS [ACL87, FRT92]. Therefore, in general, standard transactions may expect to receive poor service from real time database systems that have been designed with only real time transactions in mind. We focus here on integrating the concurrency control manager of a RTDBS catering to both real time and standard ....

P. Franaszek, J. Robinson and A. Thomasian. Concurrency Control for High Contention Environments. In ACM Trans. on Database Systems, June 1992.

....and simulated at the University of Wisconsin. In [6] Carey and Livny describes a distributed DBMS model, an extension to their centralized model. Different simulation parameters are examined through simulations. Several papers about concurrency control have also been written by Thomasian et al. [7, 12]. The most important difference between our approach and the earlier approaches, is that we focus on data shipping page server OODBs, while earlier approaches have been done in the context of query shipping relational database systems. Also, inter operation and inter transaction times are expected ....

P. A. Franaszek, J. T. Robinson, and A. Thomasian. Concurrency Control for High Contention Environments. ACM Transactions on Database Systems, 17(2), 1992.

....ID in the lock table. Given these similar recovery mechanisms and overheads, choice of SM lock management architectures should be determined based on performance criterion. 7 Related Work A number of studies have been conducted on the performance of database lock management, to name a few see [3, 24, 25]. These studies are based on analytical and simulation models, and consider uniprocessor and SD database systems. In contrast, our study has considered implementation and performance issues in SM database systems, and has focused on empirical measurements. In [23] the performance of on line ....

P. Franaszek, J. Robinson, and A. Thomasian. Concurrency Control for High Contention Environments. ACM Transactions on Database Systems, pages 304--345, June 1992.

....Between these two extremes when one needs to judiciously decide which transaction to abort. A decision based on which transaction has done more work, such as with LC, has good payoffs. 9 Two Phase Transaction Execution We have also implemented a two phased approach to transaction execution [11, 5]. The motivation for this approach is to decrease the effect of lock conflicts when combined with I O delays. Specifically, because of the delay involved in bringing the data into the buffer, a transaction holds its locks for a longer duration, resulting in increased conflicts. We briefly describe ....

Franaszek P. A., J. T. Robinson, and A. Thomasian. Concurrency Control for High Contention Environments. ACM Transactions on Database Systems, pp. 304--345, June 1992.

....example, if the conflict is detected after the granule access, the action must be abort. Several conflict resolution mechanisms have been proposed in the literature. Variants of the 2PL include no wait [TaSG85] wound wait [RoSL78] wait die [RoSL78] running priority [FrRo85] wait depth limited [FrRT92], and delayed abort [Yu90] In addition to the pure optimistic concurrency control (OCC) broadcast OCC [KuRo81] and a combination of pure and broadcast OCC [YuDL93] have been proposed. Locking with deferred blocking [YuDi93] is a scheme that combines 2PL and OCC. Multiversioning opens a wide ....

Franaszek, P. A., Robinson, J. T., and Thomasian, A., "Concurrency Control for High Contention Environments," ACM Transactions on Database Systems 17, 2, Jun 1992, pp. 304-345.

....of hints differs from ours in that their compiler is responsible for placing hints based on a static decision of when prefetches should be issued, whereas we rely on TIP to manage the scheduling of prefetches. Research presented by Franaszek, Robinson and Thomasian is close in spirit to our own [Franaszek92]. Through simulation, they demonstrated that preexecuting database transactions in order to prefetch data or pre claim locks could significantly increase throughput because it reduced effective concurrency. However, their simulations assumed that pre execution would always cause the correct data ....

P.A. Franaszek, J.T. Robinson and A. Thomasian. Concurrency control for high contention environments. ACM TODS, V 17(2), pp. 304-345, June 1992.

....phase 1 execution, the system will perform real time scheduling and then re execute the transaction to completion in phase 2. An assumption made in this re execution is that the transaction will access the same data items as it did in the first execution, a property known as Access Invariance [FRT90, FRT92] which holds for a large class of appropriate applications, as explained below in Section 4. Thus I O will not occur and real time scheduling will guarantee transaction completion, as discussed below. Finally we come to problems of unpredictability that arise from the need to provide a ....

....such as the one in [RSS] to schedule the phase 2 execution. Predictable transaction completion is guaranteed during this phase. The two phase approach draws its inspiration from the [FRT90] paper by Franaszek, Robinson and Thomasian, where the concept of access invariance was introduced. See also [FRT92]. Their paper investigated a number of two phase transaction execution schemes comparable to our own, with the aim of improving transaction throughput at high levels of concurrency. The approach was heuristic, based on the observation that during the second phase when no I Os were needed, ....

[Article contains additional citation context not shown here]

Peter A. Franaszek, John T. Robinson, and Alexander Thomasian, "Concurrency Control for High Contention Environments", ACM TODS,, 1992, pp 304-345.

....Consequently, it has been suggested to enhance two phase locking by means of transaction aborts when a specified limit for the maximum wait depth is exceeded. This idea has been elaborated in two variants that are known under the names cautious waiting [4] 44] and wait depth limitation [30] [31]. Finally, one can also conceive using global metrics like throughput and response time as indicators of data contention. A drop in throughput or an increase of response time would be interpreted as an increasing degree of data contention. In fact, it has been proposed to exert load control on the ....

.... conflict driven load control method that exploits the same knowledge as the INF method but differs from the INF method in that it serves the system entry queue in a non FIFO manner whenever this is estimated to be beneficial, ffl WDL: the wait depth limitation method proposed by Franaszek et al. [31] where the depth of a transaction waiting chain is limited to 1, ffl HALF: the half and half method proposed by Carey et al. 10] ffl MPL: a global limitation of the MPL where the limit is tuned manually by exhaustive trials for specific experiments, ffl NO: a strawman where no load control ....

P. Franaszek, J. T. Robinson and A. Thomasian. Concurrency Control for High Contention Environments. ACM Transactions on Database Systems, 17 (2), 304-345 (1992).

No context found.

Peter A. Franaszek, John T. Robinson, and Alexander Thomasian. Concurrency control for high contention environments. ACM TODS, 17:304--345, June 1992.

No context found.

P. A. Franaszek, J. T. Robinson, and A. Thomasian. Concurrency control for high contention environments. ACM Trans. on Database Systems, 17(2):304--345, June 1992.

No context found.

FRRT92 Franaszek P. A., Robinson J. T. and Thomasian A., Concurrency Control for High Contention Environments, ACM Transactions on Database Systems 17/2, Jun 1992.

No context found.

P. A. Franaszek, J. T. Robinson, and A. Thomasian, "Concurrency Control for High Contention Environments ", ACM Transactions on Database Systems, pp. 47-55, 1992.

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