J. R. Haritsa, M. Livny, and M. J. Carey. Earliest deadline scheduling for real-time database systems. In Proc. of Real-Time Systems Symposium, pages 232--242, 1991.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....prices. These applications also have the need for transaction semantics, hence real time databases are good choices. Most real time database research has evaluated their new ideas via simulation. Also, in much of the past work the only real time issue addressed is that transactions have deadlines [1,2,3,4,5,6,7]. In general, though, transactions that process real time data must use timely and relatively consistent data in order to have correct results. Data read by a transaction must be valid when the transaction completes, which leads to another constraint on completion time, in addition to a ....

J.R. Haritsa, M.J. Carey and M. Livny, "Earliest Dead-line Scheduling for Real-Time Database Systems," Proceedings of the Real-Time Systems Symposium, pp.232-242, December 1991.

....transparently. Under the partitioning approach, if a processor fails, then all tasks running on that processor need to re assigned. Such a re assignment may cause overloads on other processors, even when total utilization is less than M 1. EDF has been shown to perform poorly under overload [15, 21]. Hence, additional mechanisms are required to maintain system stability under partitioning. 5.5 The Supertasking Approach In [29] Moir and Ramamurthy observed that the migration assumptions underlying Pfair scheduling may be problematic. Specifically, tasks that communicate with external ....

J. Haritsa, M. Livny, and M. Carey. Earliest deadline scheduling for real-time database systems. In Proc. of the 12th IEEE Real-Time Systems Symposium, pages 232--243, 1991.

....QoS curve steadily declines from its maximum and eventually drops below the slopeslack technique at about 260 partitions. 7 Related Work There has been extensive research on scheduling tasks under real time performance expectations both in operating systems [14, 16, 17, 20] and database systems [3, 11, 12, 18, 19]. To the best of our knowledge, Aurora s scheduling approach that combines priority assignment and dynamic scheduling plan consauction is the first comprehensive proposal for scheduling continuous queries over real time data streams and QoS expectations. Our solutions no doubt borrow a lot from ....

....paper but can be found in [8, 21] that is initiated when an overload situation is detected and that selectively sheds load to get rid of excess load in a way that least degrades the QoS. This allows our scheduling algorithm to focus only on undefioad situations. We note here that Haritsa et al. [12] proposed an extension of EDF that is designed to handle overloads through adaptive admission control. Real time database systems [3, 11, 12, 15, 18, 19] attempt to satisfy deadlines associates with each incoming transaction, with the goal of minimizing the number of transactions that miss their ....

[Article contains additional citation context not shown here]

J.R. Haritsa, M. Livny, and M. J. Carey. Earliest Deadline Scheduling for Real-Time Database Systems. In IEEE Real-Time Systems Symposium, 1991.

....data object, and the dynamic behavior of the system. The RODAIN scheduling algorithm, called FN EDF, is designed to support simultaneous execution of both firm real time and non real time transactions. In RODAIN firm deadline transactions are scheduled according to the EDF scheduling policy [13, 12, 26]. The FN EDF algorithm guarantees that non real time transactions receive a respecified amount of execution time. The FN EDF algorithm periodically samples the execution times of all transactions. The operating system scheduling priority of a non real time transaction is adjusted if its fraction ....

Haritsa, J. R., Livny, M., and Carey, M. J. (1991). Earliest deadline scheduling for real-time database systems. In Proceedings of the 12th Real-Time Symposium, pages 232--242, IEEE Computer Society Press.

....a higher priority value to a transaction with an earlier deadline. Although EDF has been shown to improve the average success ratio of the system (fraction of transactions completing successfully within their deadline) it discriminates against longer transactions under overload conditions [Har 91, Pang 92, Pang1 92] We propose priority assignment techniques that consider the biased nature of EDF, and attempt to eliminate the discriminatory behavior by adjusting the priorities using execution history information. Two basic pieces of data that is kept in transaction history logs are the ....

....of completion for a transaction within a given time quota is estimated using the completion time history (mean and variance) and transactions with lower completion probabilities are assigned higher priority values to adjust the bias towards short transactions. Haritsa et al. points out [Har 91] the need for load control in RTDBMSs. EDF policy, by assigning higher priority values to transactions with earlier deadlines, determines the execution order of transactions. However, under 96 overload conditions, the performance of the system degrades due to assigning higher priorities to ....

[Article contains additional citation context not shown here]

Jayant R. Haritsa, Miron Livny, Micheal J. Carey, "Earliest Deadline Scheduling for Real-Time Database Systems", Proc. of the IEEE Real-Time Systems Symposium, 1991.

....a primary mechanism to adjust resource allocation in the absence of a priori knowledge of resource supply and demand. This is in contrast to early optimization based QoS adaptation techniques that assumed accurate models of application resource requirements. Examples of such approaches include [4][7]. In [7] a transaction scheduler called AED monitors the system deadline miss ratio and adjusts task priorities to improve the performance of EDF in overload. The DQM algorithm [4] features a feedback mechanism that changes task QoS levels according to the sampled CPU utilization or deadline ....

....mechanism to adjust resource allocation in the absence of a priori knowledge of resource supply and demand. This is in contrast to early optimization based QoS adaptation techniques that assumed accurate models of application resource requirements. Examples of such approaches include [4] 7] In [7], a transaction scheduler called AED monitors the system deadline miss ratio and adjusts task priorities to improve the performance of EDF in overload. The DQM algorithm [4] features a feedback mechanism that changes task QoS levels according to the sampled CPU utilization or deadline misses. ....

[Article contains additional citation context not shown here]

J.R. Haritsa, M. Livny and M. J. Carey. Earliest Deadline Scheduling for Real-Time Database Systems. IEEE RealTime Systems Symposium, 1991.

....for RM (CPU Utilization 0.88) by studying the average case and the harmonic characteristics of tasks. However, these algorithms do not allow for the possibility of variable execution and are unable to cope with overload conditions effectively. As a result, many besteffort scheduling heuristics [17][18] have been proposed in an attempt to cope with overload conditions. A related school of thought is the study of imprecise computation technique [19] 20] A task is divided into two parts, a mandatory sub task which must be executed and an optional sub task which serves to refine and improve ....

J. R. Haritsa, M. Livny, and M. J. Carey, "Earliest Deadline Schedu-ling for Real-time Database Systems," In Proceedings of the IEEE Real-Time Systems Sym-posium, pp 232-243, Dec 1991.

....[Abbo90, Care89, Chen91] A number of papers on related issues have also appeared. These include work on a protocol for timed atomic commitment [Davi91] fast recovery protocols for real time database systems [Kort90b, Vrbs88] priority assignment protocols for managing an overload situation [Hari91b], and the application of imprecise computation technique [Lin87, Liu87] in realtime database systems [Hou89, Pu92, Rama91b] There are also a number of studies on the correctness notions for data consistency other than serializability that can provide a higher level of concurrency by sacrificing ....

Haritsa, J. R., M. J. Carey, and M. Livny, "Earliest Deadline Scheduling for Real-Time Database Systems," Proceedings of the 12th IEEE Real-Time Systems Symposium, December 1991.

....TMN transactions must get enough resources to complete. The RODAIN scheduling algorithm, called FNEDF, is designed to support simultaneous execution of both firm real time and non real time transactions. In RODAIN firm deadline transactions are scheduled according to the EDF scheduling policy [29, 12]. The FN EDF algorithm guarantees that non real time transactions receive a respecified amount of execution time. The FN EDF algorithm periodically samples the execution times of all transactions. The operating system scheduling priority of a non real time transaction is adjusted if its fraction ....

J. R. Haritsa, M. Livny, and M. J. Carey. Earliest deadline scheduling for real-time database systems. In Proc. of the 12th Real-Time Symp., pp 232--242, Los Alamitos, Calif., 1991. IEEE, IEEE Computer Society Press.

....than choosing to maintain optimal service to some views, thereby e ectively relieving the starvation. The EDF (Earliest Deadline First) scheduling algorithm [1] is applied in a wide range of real time systems and has been shown to be optimal with respect to di erent metrics in the literature [2, 8, 11]. However, in overload conditions, it performs very poorly because EDF gives the highest priority to those tasks that are close to missing their deadlines. Therefore, a typical phenomenon that may happen with EDF when the system is overloaded is the domino e ect , since the rst task that misses ....

....the overload interval. To the best of our knowledge, previous work [2, 14, 3, 9, 15] focused on the case when EPU is the measure of scheduling performance. Some algorithms have also been proposed to deal with overloads that improve the performance of EDF that is optimal when system is underloaded [2, 8, 11]. Baruah et al. 13] have shown that there exists an upper bound on the performance of any on line (preemptive) algorithm working in overload conditions. Traditionally, the goodness of an online algorithm working in overload conditions is measured with respect to a clairvoyant scheduler (one ....

[Article contains additional citation context not shown here]

J. R. Haritsa, M. Livny, M. J. Carey. Earliest deadline scheduling for real-time database systems. In Proceedings of Real-Time Systems Symposium, December 1991.

....work and previous work is that we distinguish the tight feasibility and the loose feasibility which is concerned by all previous work. Also we derive results for speci c (uniform or regular periods) task systems with given task start times. More related work on real time scheduling can be found in [2, 3, 5, 9, 10]. 6 Conclusions To refresh semi dynamic Web pages whose contents are derived from a source database is a important issue due to the arbitrary update patterns of base data and a large number of refresh queries scheduled on a single processor. In this paper, we view a website as a set of Web ....

J. R. Haritsa, M. Livny, M. J. Carey. Earliest deadline scheduling for real-time database systems. In Proceedings of Real-Time Systems Symposium, December 1991.

....ffl Dynamic planning based approaches: Feasibility is checked at run time, i.e. a dynamically arriving task is accepted for execution only if it found feasible. ffl Dynamic best effort approaches: The system tries to do its best to meet deadlines. In the earliest deadline first (EDF) LL73, HLC91] policy, the transaction with the earliest deadline has the highest priority. Other transactions will receive their priorities in descending deadline order. In the least slack first (LSF) AGM88a] policy, the transaction with the shortest slack time is executed first. The slack time is an ....

....in [AGM88a] 3.3 Transaction Scheduling in RODAIN The RODAIN scheduling algorithm, called FN EDF, is designed to support simultaneous execution of both firm real time and non realtime transactions. In RODAIN, firm deadline transactions are scheduled according to the EDF scheduling policy [LL73, HLC91] The FN 14 EDF algorithm guarantees that non realtime transactions will receive a prespecified amount of execution time. The FN EDF algorithm periodically samples the execution times of all transactions and of non realtime transactions. The operating system 1 scheduling priority of a ....

J. R. Haritsa, M. Livny, and M. J. Carey. Earliest deadline scheduling for realtime database systems. In Proceedings of the 12th Real-Time Symposium, pages 232--242, Los Alamitos, Calif., 1991. IEEE, IEEE Computer Society Press.

....be enhanced to allow for the deletion of specified tasks, such deletion corresponding to the task being moved to a Gprocessor and hence no longer a contender for the L processor. much larger than in, e.g. EDF scheduling. Given the unacceptably poor behavior of EDF under conditions of overload [11,12],however, this seems to be an unavoidable price to pay in order to obtain graceful degradation upon overload. Note that once a task has migrated to a Gprocessor in Algorithm kOnLine, it is guaranteed to complete with no further preemptions# hence, Algorithm kOnLine tends to execute tasks with ....

J. Haritsa, M. Carey, and M. Livny. Earliest-deadline scheduling for real-time database systems. In Proceedings of the Real-Time Systems Symposium, 1991.

....of the above is discussed in detail. 2.1 PARTITIONED AND PARALLEL LOGGING AND RECOVERY The following example motivates the need for partitioning the log in RTDBs. In RTDBs, usually, cpu scheduling, i o scheduling and concurrency control conflict resolution are based on transaction priorities ([10, 11, 12, 13]) But, conventional (non partitioned) logging and recovery techniques are sequential activities that do not work well in such a priority oriented and preemptive RTDB setting. The following example illustrates that it is difficult to attain the key (last 112 REAL TIME DATABASE SYSTEMS: ....

J. R. Haritsa, M. J. Carey, and M. Livny, "Earliest deadline scheduling for real-time database systems," in Proceedings of the Real-Time Systems Symposium, Dec 1991.

No context found.

J. Haritsa, M. Carey, and M. Livny, "Earliest-Deadline Scheduling for Real-Time Database Systems," Proc. 12th IEEE Real-Time Systems Symp., San Antonio, Tex., Dec. 1991.

....the system starts to miss the deadlines of periodic updates since sporadic updates have higher priority than periodic updates. To minimize the number of missed deadlines under these overload conditions, we propose to use in MANDATE the Adaptive Earliest Deadline scheduling policy described in [16], which has been shown to provide robust good performance under a variety of workloads and system loading conditions. 7 CONCLUSIONS The Management Information Base (MIB) of a network management system is a critical component since it provides the interface between all functions of the network ....

J. Haritsa, M. Livny, and M. Carey. Earliest Deadline Scheduling for Real-Time Database Systems. In Proc. IEEE Real-Time Systems Symposium, December 1991.

.... tasks in deadline order, is optimal for uniprocessor systems under non overloaded conditions in the sense that it meets all deadlines whenever it is feasible to do so [3]# however, under conditions of overload, it has been experimentally observed to perform worse than even random scheduling [4]. System Model: We address here the problem of scheduling a number of tasks on a single processor . The processor is assumed to be fully preemptable, in that no penalty is incurred if a task currently executing on the processor is interrupted, and resumed at a later time. EachtaskT is completely ....

J. Haritsa, M. Carey, and M. Livny. Earliest-deadline scheduling for real-time database systems. In Proceedings of the Real-Time Systems Symposium, 1991.

No context found.

J. R. Haritsa, M. Livny, and M. J. Carey. Earliest deadline scheduling for real-time database systems. In Proc. of Real-Time Systems Symposium, pages 232--242, 1991.

No context found.

J. Haritsa, M. Livny, and M. Carey. Earliest deadline scheduling for real-time database systems. In 12th RealTime Systems Symposium, Dec 1991.

No context found.

Haritsa J, Carey MJ, Livny M. Earliest deadline scheduling for real-time database systems. Proceedings of the Real-Time Systems Symposium, 1991.

No context found.

Jayant R. Haritsa, Miron Livny, and Michael J. Carey. Earliest deadline scheduling for real-time database systems. In Proceedings of the 12th Real-Time Systems Symposium, December 1991.

No context found.

J. Haritsa, M. Livny, M. Carey. Earliest deadline scheduling for real-time database systems, in: Proceedings of the 12th IEEE Real-Time Systems Symposium, 1991.

No context found.

J. Haritsa, M. Livny, and M. Carey. Earliest Deadline Scheduling for Real-Time Database Systems. In Procs. IEEE Real-Time Systems Symposium, pages 232--242, 1991.

No context found.

J.R. Haritsa, M. Livny, and M.J. Carey, "Earliest Deadline Scheduling for Real-Time Database Systems," Proceedings of Real-Time Systems Symposium, December 1991, pp. 232-242.

No context found.

Haritsa, J.R., Carey M.J. & Livny, M., "Earliest Deadline Scheduling for Real-Time Database Systems", in Proceedings of Real-Time Systems Symposium, pp. 232-242, 1991.

First 50 documents  Next 50

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