S. Chen , J.A. Stankovic, J.F. Kurose and D. Towsley, "Performance evaluation of two new disk scheduling algorithms for real-time systems", Journal of Real Time Systems 3(3), pp. 306-336 (1991).  Home/Search   Document Details and Download   Summary   Related Articles   Check

....contract number E30602 97 2 0287 and in part under contract number F30602 96 1 0160. Mr. Molano was funded by a research grant from the Community of Madrid and by the National R D Program of Spain under contracts TIC96 0982 and TIC97 0438. one type (namely network packets) Work described in [1, 2, 7, 8] study various aspects of real time disks and filesystems. The approach used in [1] is a variation of the Scan algorithm in which the SCAN direction changes towards the Earliest Deadline First request only if the deadline is considered to be met. At any scheduling point the request with the ....

....of the Scan algorithm in which the SCAN direction changes towards the Earliest Deadline First request only if the deadline is considered to be met. At any scheduling point the request with the Earliest Feasible deadline is chosen for service and the disk head scans towards it. Chen et al. [7] propose two algorithms which combine deadline information and disk service time information. They show that these algorithms minimize transaction loss ratio in real time database applications. Both [1] and [7] are proposed and studied under soft deadline schemes. A scheduling model close to ours ....

[Article contains additional citation context not shown here]

S. Chen, J. A. Stankovic, J. F. Kurose, D. Towsley. "Performance Evaluation of Two New Disk Scheduling Algorithms for Real-Time Systems". The Real-Time Systems Journal 3 (1991), 307-336.

.... algorithm such as the Shortest Seek Time First or SCAN (Elevator) orders the sequence of IO requests to minimize the average disk head seek time [16] On the other hand, the disk scheduler in a RTDBS primarily concerns the timing constraints of transactions in processing data access requests [3, 11, 13, 37]. As discussed in [49] one important issue in scheduling the IO requests of a RTDB transaction is the assignment of individual deadlines to the requests on the basis of the transaction deadline. This issue has not been addressed explicitly by the recent work performed on IO scheduling. We ....

....algorithm. Although using the priority based disk scheduling algorithms described above help IO requests meet their timing constraints, the overall IO performance in terms of the average seek time can become worse since some requests can receive very poor service. The algorithms provided in [13] attempt to reduce the overall seek time while taking the timing constraints into account in servicing the requests. In assigning request priorities, both the location and the deadline of requests are considered. A request very close to the disk arm can be assigned a high priority even if it has a ....

S. Chen, J.A. Stankovic, J. Kurose, D. Towsley `Performance Evaluation of Two New Disk Scheduling Algorithms for Real-Time Systems', Real-Time Systems, vol.3, no.3, 1991, pp.307336.

....policies based on enforcing data consistency by using a two phase locking concurrency control mechanism. An extended version of their work appeared recently in [4] In [3] they provided a study of various algorithms for scheduling IO requests with deadlines. Carey et al. 11] and Chen et al. [13] also discussed some new approaches to priority based IO scheduling. In [35] and [36] Sha et al. presented a new priority based concurrency control protocol called priority ceiling. The performance of this protocol was examined in [37] by using simulations. In [23] Huang et al. developed and ....

S.Chen, J.A.Stankovic, J.Kurose, D.Townley, "Performance Evaluation of Two New Disk Scheduling Algorithms for Real-Time Systems", Real-Time Systems Journal, vol.3, pp.307336, 1991.

....usually are considered to be trade secrets. This section will discuss a few success stories. 4. 1 Disks Disk Scheduling Numerous studies have investigated the performance of disk I O scheduling algorithms involving command queuing (e.g. 18, 51, 52, 63, 64, 65] real time applications (e.g. [66, 67, 68, 69]) and database applications (e.g. 70] Event driven simulation studies of disk scheduling algorithms that use rotational position information appear in [52] these studies were used by Hewlett Packard s Disk Memory Division in the design of products. Disk Cache Performance evaluation methods ....

S. Chen, J. F. Kurose, J. A. Stankovic, and D. Towsley, \Performance evaluation of two new disk scheduling algorithms for real-time systems," Tech. Rep. UMCS -1990-077, Department of Computer Science, University of Massachusetts at Amherst, 1990.

....drawback of such a scheme is that the service cycle needs to be large enough if the advantages of SCAN ordering have to be realised. Large service cycle implies that only request streams with coarse granularity of deadlines can be scheduled. Other algorithms for deadline based I O are described in [7, 8, 9]. All of them use EDF schedule as the basic scheme and reorder requests so as to reduce seek and rotational latency overhead. 1 To achieve a performance level as close to the SCAN algorithm as possible while meeting all disk requests deadlines, we propose a Deadline Sensitive SCAN Algorithm ....

S.Chen, J.A.Stankovic, J.F.Kurose, and D.Towsley, \Performance Evaluation of Two New Disk Scheduling Algorithms for Real-Time Systems", Journal of Real-Time Systems, Vol. 3, 307-336, 1991. 6

....are serviced one at a time at the disk while maintaining the timing requirements of the accesses. Many schemes for multi plexing time critical accesses to I O devices like Round Robin(RR) SCAN, Earliest deadline first, etc. have been previously studied in [Abbott and Garcia Molina, 1990] [Chen et al. 1991]. The high retrieval latency and the non preemptible nature of disks makes strategies like RR and SCAN attractive for real time, high volume I O accesses as those found in CM. The RR scheme amortizes latency over large data fetches while the SCAN algorithm minimizes latency by re ordering accesses ....

..... k (number of additional frames fetched in each cycle) Figure 3.8: Slack vs. k. The problem of scheduling accesses to I O systems has been studied extensively as early as [Frank, 1969] Teorey and Pinkerton, 1972] and [Fuller, 1975] More recently, Abbott and Garcia Molina, 1990] and [Chen et al. 1991] have addressed the problem in the context of real time systems. With the growing interest in continuous media server design, recent work by [Gemmell, 1993] Reddy and Wyllie, 1993] Chen et al. 1993] Anderson et al. 1992] Rangan and Vin, 1993a] Gemmell and Christodoulakis, 1992] ....

Shenze Chen, James Kurose, and Don Towsley. Performance Eval- uation of Two New Disk Scheduling Algorithms. Real Time Systems, September 1991.

....take into account any deadline constraints on the I O requests. The simplest algorithm for deadline based scheduling is EDF [15] However it does not take into account the relative positions of requested data on the disk. Hybrids of SCAN and EDF algorithms for deadline based I O are described in [1, 6, 7, 21]. All of them use EDF schedule as the basic scheme and reorder requests so as to reduce seek and rotational latency overhead. To achieve a performance level as close to the SCAN algorithm as possible while meeting all disk requests deadlines, IRS uses a Deadline Sensitive SCAN Algorithm ....

S. Chen, et. al., \Performance Evaluation of Two New Disk Scheduling Algorithms for Real-time Systems ", Journal of Real-time Systems, Vol. 3, 307336, 1991.

....incurs significant seek time and rotational latency overhead. This limitation has been addressed by several disk scheduling algorithms, including Priority SCAN (PSCAN) Earliest Deadline SCAN, Feasible Deadline SCAN (FD SCAN) SCAN EDF, Shortest Seek Earliest Deadline by Order Value (SSEDO, SSEDV) [1, 4, 5, 15], etc. These algorithms reduce seek and rotational latency overhead by starting from an EDF schedule and reorder requests so as to tend towards a SCAN schedule without violating request deadlines. Unlike the systems for which these scheduling algorithms were designed, today s general purpose file ....

....through virtual environments yield real time requests with low average response time requirements. EDF and fixed priority schedulers are suitable for hard realtime applications [14] while scheduling algorithms such as FD SCAN and SSEDV SSEDO are are suitable for soft real time applications [1, 5]. Just in time schedulers (which schedule requests just prior to their deadlines) are desirable for real time applications that initiate data consumption at deadlines (e.g. video playback) Finally, algorithms that schedule requests at the earliest possible instants prior to their deadlines, and ....

S. Chen, J. A. Stankovic, J. F. Kurose, and D. Towsley. Performance Evaluation of Two New Disk Scheduling Algorithms for Real-Time Systems. Journal of Real-Time Systems, 3:307--336, 1991.

....[29] Our experiments have demonstrated how the combination heuristic brings about simultaneous improvement of both contract adherence and performance. Real time disk schedulers (either pure or in combination with seek reducing schedulers) are commonly used to serve and view multimedia content [9, 11]. Under certain circumstances, it is possible for deceptive idleness to cause such schedulers to multiplex between requests from di erent processes, and consistently violate deadlines. Webelievethat the anticipatory scheduling framework is applicable to realtime scheduling, but a full exploration ....

S.Chen,J.A.Stankovic, J. F. Kurose, and D. Towsley.Performance evaluation of twonewdisk scheduling algorithms for real-time systems. Journal of Real-Time Systems, 3(3):307-336, Sept. 1991.

....of this paper appeared in the Proceedings of the ACM SIGMETRICS 98 conference, Madison, WI, pages 44 55, June 1998. y Corresponding author 1 (PSCAN) Earliest Deadline SCAN, Feasible Deadline SCAN (FD SCAN) SCAN EDF, Shortest Seek Earliest Deadline by Order Value (SSEDO, SSEDV) etc. [1, 7, 8, 21]. These algorithms start from an EDF schedule and reorder requests so as to reduce the seek and rotational latency overhead without violating request deadlines. Unlike the systems for which these scheduling algorithms were designed, today s general purpose file and operating systems ....

....through virtual environments yield real time requests with low average response time requirements. EDF and fixed priority schedulers are suitable for hard real time applications [18] while scheduling algorithms such as FD SCAN and SSEDV SSEDO are are suitable for soft real time applications [1, 8]. Just in time schedulers (which schedule requests just prior to their deadlines) are desirable for real time applications that initiate data consumption at deadlines (e.g. video playback) Finally, algorithms that schedule requests at the earliest possible instants prior to their deadlines, and ....

S. Chen, J. A. Stankovic, J. F. Kurose, and D. Towsley. Performance Evaluation of Two New Disk Scheduling Algorithms for Real-Time Systems. Journal of Real-Time Systems, 3:307--336, 1991.

....[29] Our experiments have demonstrated how the combination heuristic brings about simultaneous improvement of both contract adherence and performance. Real time disk schedulers (either pure or in combination with seek reducing schedulers) are commonly used to serve and view multimedia content [9, 11]. Under certain circumstances, it is possible for deceptive idleness to cause such schedulers to multiplex between requests from di erent processes, and consistently violate deadlines. We believe that the anticipatory scheduling framework is applicable to realtime scheduling, but a full ....

S. Chen, J. A. Stankovic, J. F. Kurose, and D. Towsley. Performance evaluation of two new disk scheduling algorithms for real-time systems. Journal of Real-Time Systems, 3(3):307-336, Sept. 1991.

.... and data conflict resolution [Abbo88, Abbo89, Agra92, Buch89, Hari90, Hari90b, Huan89, Huan91b, Huan91c, Lee93, Lee93b, Lee93c, Lee93d, Lin90, Sha88, Son90, Son90b, Song90, You93] processing of queries with real time constraints [Hou89] buffer management [Care89, Huan90] and I O scheduling [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 ....

Chen, S., J. A. Stankovic, J. Kurose, and D. Towsley, "Performance Evaluation of Two New Disk Scheduling Algorithms for Real-Time Systems," Journal of Real-Time Systems, September, 1991.

....conserving. Nevertheless, the non real time results discussed above have been extended by several researchers to heuristic policies based on the ED result. Much existing work stems from the combinatorial optimization approach, but such formulations lead to complex analysis. Kurose et al. in [9] evaluated in a testbed the performance of SSEDO (Shortest Seek Earliest Deadline by Ordering) and SSEDV (Shortest Seek Earliest Deadline by Value) Both policies sort requests by deadline, but SSEDO uses also a weighted seek distance and SSEDV checks whether deadlines are feasible prior to ....

S. Chen, J. A. Stankovic, J. F. Kurose, and D. Towsley, "Performance Evaluation of Two New Disk Scheduling Algorithms for Real-Time Systems," Journal of Real-Time Systems, vol. 3, pp. 307--336, 1991.

....disk requests. Disk scheduling algorithms can be broadly divided into two classes: those optimized to service best effort requests (e.g. SCAN, Shortest Access Time First (SATF) 38, 17, 33] and those optimized to service real time requests (e.g. Earliest Deadline SCAN, Feasible Deadline SCAN [1, 5, 31, 46]) Whereas the former class of algorithms attempts to improve the performance of best effort requests by reducing disk seek and rotational latency overheads, the latter class attempts to meet deadlines of real time requests while reducing disk overheads. In homogeneous environments, depending on ....

S. Chen, J. A. Stankovic, J. F. Kurose, and D. Towsley. Performance Evaluation of Two New Disk Scheduling Algorithms for Real-Time Systems. Journal of Real-Time Systems, 3:307--336, 1991.

.... been much research in scheduling algorithms for real time applications: priority FCFS [Abbott89] EDF (Earliest Deadline First) P SCAN (Priority SCAN) FDSCAN (Feasible Deadline SCAN) SSEDO (Shortest Seek and Earliest Deadline by Ordering) SSEDV (Shortest Seek and Earliest Deadline by Value) [Chen90], Greedy Priority Inversion [Li95] and Preempted Missed Only (PMO) Chen94] There has also been research in scheduling algorithms 40 for database applications (Priority based scheduling [Carey89] and in distributed scheduling [Worthington95a] We restrict our attention to FCFS, CLOOK, and ....

Shenze Chen, James F. Kurose, John A. Stankovic, and Don Towsley. Performance evaluation of two new disk scheduling algorithms for real-time systems. Technical report UM-CS-1990-077. University of Massachusetts at Amherst, 1990.

....executes, subsequent computations and disk accesses of the transaction will not be known. Also, disk access delays constitute one of the primary sources of unpredictability of transaction executions. Naturally, the approach to I O scheduling should also be deadline sensitive (see for example [6]) We point out that our approach keeps track of deadlines during the prefetch phase and will abort a transaction if it is already past its deadline. This gives us an important advantage in situations where deadlines are much too small. As mentioned earlier, we might be able to provide a ....

S. Chen, J. Stankovic, J. Kurose, and D. Towsley, "Performance Evaluation of Two New Disk Scheduling Algorithms for Real-Time Systems", Real-Time Systems, Sept. 1991.

.... other high priority transactions (due to write locks) 119] SCAN algorithms can also be revised using the knowledge of prioritized groups of disk requests [33] Another approach tries to balance time constraints of disk requests and the overall I O performance in terms of the average seek time [37]. 7 Conclusions and Future Work Temporal database research has been active for about twenty years. Initially the focus was on temporal relational databases. A decade ago research on temporal object oriented data models was launched under the guise of versions in an engineering database [157] ....

Chen, S., Stankovic, J. A., Kurose, J. and Townley, D., "Performance Evaluation of Two New Disk Scheduling Algorithms for Real-Time Systems," Journal of Real-Time Systems 3, pp. 307--337, 1991.

....meeting transaction deadline is the main concern on protocol performance, we employ the earliest deadline first (EDF) policy for transaction priority assignment. For studies on other priority assignment policies based on transaction slack time, criticalness, or length, the reader is referred to [1, 6, 4, 8]. Besides PI, PA and CP discussed in Section 2, we also look at two other policies, NRT (non real time) and WAIT, for the sake of performance comparisons. Representing a non real time transaction processing system, NRT is the performance baseline in these experiments. Under NRT, transactions are ....

....wait oriented twophase locking approach. 15 last stage of transaction execution, is usually relatively short, priority inheritance may be applied to the validating transaction which blocks a higher priority transaction. We may futher extend this work by incorporating real time I O scheduling [5, 2, 4, 9]. It would be interesting to re examine the various schemes studied in this paper in a system that integrates both priority based CPU scheduling and real time I O scheduling. ....

Chen, S., J. Stankovic, J. Kurose, and D. Towsley, "Performance Evaluation of Two New Disk Scheduling Algorithms for Real-Time Systems," A Technical Report, COINS 90-77, University of Massachusetts, Aug. 1990.

.... requirements but also operate under timing constraints [36, 37, 50, 69] The algorithms and protocols that must be integrated 23 include: cpu scheduling, concurrency control, conflict resolution, transaction restart, transaction wakeup, deadlock, buffer management, and disk I O scheduling [21, 23, 25, 26, 50, 100]. Each of these algorithms or protocols should directly address the real time constraints. To date, work on real time databases has investigated a centralized, secondary storage real time database [1, 2, 3] As is usually required in traditional database systems, work so far has required that all ....

....ffl In a real time system, I O scheduling is an important issue with respect to the system performance. In order to minimize transaction loss probability, a good disk scheduling algorithm should take into account not only the time constraint of a transaction, but also the disk service time [25]. ffl Used for I O, the earliest deadline discipline ignores the characteristics of disk service time, and, therefore, does not perform well except when the I O load is low. ffl Various conflict resolution protocols which directly address deadlines and criticalness can have a important impact on ....

S. Chen, J. Stankovic, J. Kurose, and D. Towsley, "Performance Evaluation of Two New Disk Scheduling Algorithms for Real-Time Systems," Real-Time Systems, Vol. 3, No. 3, September 1991.

No context found.

Chen S., Stankovic, J. A., Kurose, J. F., and Towsley, D., "Performance Evaluation of Two New Disk Scheduling Algorithms for Real-Time Systems," COINS Tech. Report 90-77, UMass, Aug. 1990.

....We would like to know in an I O bound environment how the I O scheduling affects the system performance as compared 16 to the buffer allocation scheme. Unfortunately, we can not explore this interesting issue due to limitations on the physical testbed. A recent study based on simulation [2] has demonstrated that I O scheduling is a key processing component in real time database systems. It is likely that I O scheduling is more important than the buffer allocation. The last factor we want to point out is the buffer contention. Our discussion on buffer allocation has focused on the ....

Chen, S., J. Stankovic, J. Kurose, and D. Towsley, "Performance Evaluation of Two New Disk Scheduling Algorithms for Real-Time Systems," Submitted for publication, August, 1990.

....be inappropriate for real time systems, the use of nontime cognizant disk scheduling algorithms are not appropriate for scheduling disk requests. It has been shown that disk scheduling algorithms that combine a scan and deadline requirement works considerably better than conventional algorithms [4]. It is likely that some combined solution will prevail where critical data is placed and pinned in (non volatile) main memory, and other, less critical data is stored on the disk using time cognizant disk scheduling. A temporal database is a real time database. Although both temporal ....

S. Chen, J. Stankovic, J. Kurose, and D. Towsley, Performance Evaluation of Two New Disk Scheduling Algorithms for Real-Time Systems, Real-Time Systems Journal, Vol. 3, No. 3, September 1991.

No context found.

S. Chen , J.A. Stankovic, J.F. Kurose and D. Towsley, "Performance evaluation of two new disk scheduling algorithms for real-time systems", Journal of Real Time Systems 3(3), pp. 306-336 (1991).

No context found.

S. Chen, J. A. Stankovic, J. F. Kurose, and D. Towsley. Performance evaluation of two new disk scheduling algorithms for real-time systems. Technical Report UM-CS-1990-077, University of Massachusetts, Amherst, 1991.

No context found.

S. Chen, J. A. Stankovic, J. F. Kurose, and D. Towsley. Performance evaluation of two new disk scheduling algorithms for real-time systems. Technical Report UM-CS-1990-077, 1990.

First 50 documents

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