P.S. Yu, M-S. Chen, and D.D. Kandlur. Grouped sweeping scheduling for DASD-based multimedia storage management. Multimedia Systems, 1(1):99--109, January 1993.  Home/Search   Document Not in Database   Summary   Related Articles   Check

...., as a parameter important to the overall framework. Section 2.2 derives the optimal value of that minimizes system cost. We use this to detail CBS concisely. tion simplifies the discussion about the CBS framework. Extensions of CBS to scheduling techniques such as Group Sweeping Scheme [YCK93] that require 2 buffers on behalf of each display is a straightforward extension of this work. Database Size Arrival Rate Access Distribution Exp. Buffer Req. Exp. Disk Bandwidth Req. System Utilization Factor Size of Buffer Pool Number of Disk Streams System Generator Buffer Mgt. Technique ....

.... admission control component that monitors and controls available resources. When resources (memory and disk bandwidth) are exhausted, it queues newly arrived requests in order to prevent hiccups. A number of studies have outlined the design and implementation of such a component [GM98, GZS 97, YCK93, OBRS94, WSY95, TPBG93] This section begins with and how to derive its optimal value in order to minimize system cost. Next, we present BMDT, followed by a discussion of the configuration planner. For each, we describe the impact of the VCR operations. The configuration planner and BMDT are ....

P.S. Yu, M-S. Chen, and D.D. Kandlur. Grouped sweeping scheduling for DASD-based multimedia storage management. Multimedia Systems, 1(1):99--109, January 1993.

....is served by N nodes concurrently, each node only needs to serve a bitrate of R, N for each video stream. Many existing video server designs employ round based schedulers such as SCAN and its variants [1] In our design, we 234 employ the Grouped Sweeping Scheme (GSS) proposed by Yu, et al. [2] to schedule a node s disk retrieval and network transmission. Compared to the more common SCAN scheduler that maximizes throughput at the expense of buffer overhead, GSS allows one to control the tradeoff between disk efficiency and buffer requirement. Tiffs is a crucial feature as disk ....

....in a micro round using the SCAN scheduler. These retrieved data blocks will then be transmitted at a rate of RJN for a duration of T seconds, which precisely equals to g micro rounds, Therefore the g groups are eftictively staggered in time and this reduces the combined buffer requirement [2]. 3.3 Fault Tolerauce In a server less VoD system, fault tolerance becomes an essential capability as reliability of STBs and PCs will be significantly lower than dedicated video servers located in a data centre run by professional operators around the clock. Moreover, given the relatively large ....

P.S. Yu, M. S. Chen, and D. D. Kandlur, "Grouped Sweeping Scheduling for DASD-based Multimedia Storage Management", A CM Multimedia Systems, vol. 1 (2), 1993, pp.99-109.

....This translates to a rapidly increasing DRAM buffering cost. A large DRAM buffer is especially necessary for servers which stream to a large number of clients. Multimedia server architects have tried to cope with this performance gap by proposing solutions ranging from simple resource trade offs [13, 25] to more complex ones that require substantial engineering effort [2, 7, 10] Micro electro mechanical systems (MEMS) based storage is an emerging technology that promises to bridge the performance gap between magnetic disks and DRAM [1] MEMS devices are predicted to be an order of magnitude ....

P. Yu, M.-S. Chen, and D. Kandlur. Grouped sweeping scheduling for DASD-based multimedia storage management. Multimedia Systems, 1(1):99--109, January 1993. 12

....This translates to a rapidly increasing DRAM bu#ering cost. A large DRAM bu#er is especially necessary for servers which stream to a large number of clients. Multimedia server architects have tried to cope with this performance gap by proposing solutions ranging from simple resource trade o#s [13, 24] to more complex ones that require substantial engineering e#ort [2, 7, 10] Micro electro mechanical systems (MEMS) based storage is an emerging technology that promises to bridge the performance gap between magnetic disks and DRAM [1] MEMS devices are predicted to be # This research was ....

P. Yu, M.-S. Chen, and D. Kandlur. Grouped sweeping scheduling for DASD-based multimedia storage management. Multimedia Systems, 1(1):99--109, January 1993. 24

....to playback the data in one bu er. In our simulation a bu er stores a whole disk track, and the playback rate is assumed to be 150 KB sec. Disk head movement is based on the Grouped Sweeping Scheduling paper.tex; 25 05 2001; 16:29; p. 12 Delivery Scheduling of Multimedia Streams 13 (GSS) algorithm [27], where the number of groups is one. The amount of extra time in each disk service round after delivering one track for each stream gives us the disk bandwidth utilization percentage. The average disk bandwidth utilization is the statistical average from all the disk service rounds, and the ....

P. S. Yu, M. S. Chen, and D. D. Kandlur, \Grouped Sweeping Scheduling for DASD-based Multimedia Storage Management," Multimedia Systems, Vol. 1, No. 3, pp. 99-109,

....time all increase with the number of servers in the system. Therefore, the scalability of the system will be limited by either the economy of memory buffers or the tolerance of the system response time by the user. In this section, we propose an extension of the grouped sweeping scheme (GSS) [12], called asynchronous group sweeping scheme to substantially reduce server buffer requirement and scheduling delay. A. Extending the Grouped Sweeping Scheme The original GSS algorithm in [12] is designed for scheduling retrieval requests in a magnetic disk. The traditional first in, first out ....

....response time by the user. In this section, we propose an extension of the grouped sweeping scheme (GSS) 12] called asynchronous group sweeping scheme to substantially reduce server buffer requirement and scheduling delay. A. Extending the Grouped Sweeping Scheme The original GSS algorithm in [12] is designed for scheduling retrieval requests in a magnetic disk. The traditional first in, first out scheduling algorithm has poor disk utilization in continuous media applications [8] 9] because in the worst case, the disk arm may need to seek back and forth between the innermost track and ....

[Article contains additional citation context not shown here]

P. S. Yu, M. S. Chen, and D. D. Kandlur, "Grouped sweeping scheduling for DASD-based multimedia storage management," ACM Multimedia Syst., vol. 1, pp. 99--109, 1993.

.... were analyzed in [14, 15] Non contiguous disk allocation of streams was proposed and admission control based on constrained layout designs were analyzed in [12] A disk head scheduling algorithm, group sweeping scheme with the objective of minimizing buffer space and access time was presented in [20]. For the SPIFFI video server, 9] proposed a real time priority based disk scheduling algorithm that partitions requests into priority classes based on the nearness of their deadlines. These algorithms deal with scheduling a single disk. For multiple disk parallel disk arrays, simple reading of ....

P. S. Yu, M-S. Chen, and D. D. Kandlur. Grouped Sweeping Scheduling for DASD-based Multimedia Storage Management. Multimedia Systems, 1(1):99--109, January 1993.

....consumer may experience glitches in the delivery of the movie. To ensure glitch free service, the video server has to guarantee finishing the three components of the service in a fixed amount of time. Guaranteeing delay bounds in service component (1) is addressed by appropriate disk scheduling [2, 1, 3, 4]. The problem of ensuring delay bounds in service component (2) is addressed in this paper. When multiple transmissions take place simultaneously over the network, the delays experienced by individual transmissions depend on the contention experieinced in the network. Worst case assumptions of ....

....links and this contention varies based on the network load. Also, a network transfer requires multiple resources (links, input and output ports) unlike the assumption of requiring only one resource in these studies. Hence, these results cannot be directly applied to our problem. Recent work [2, 1, 3, 4, 8] has looked at disk scheduling in a video server. File systems for handling continuous media have been proposed [9, 10, 11, 12, 13] Multicomputer based video servers are studied in [14, 15, 16] Related work in multicomputer communication includes estimation of delays in the network [17, 18, 19] ....

P. S. Yu, M. S. Chen, and D. D. Kandlur. Grouped sweeping scheduling for dasd-based multimedia storage management. Multimedia Systems, 1:99--109, 1993.

....of each file i, denoted by p i , be fixed. Intuitively, the minimum p i is p min i : minfnjnR i g, and any larger p i will demand more disk throughput than necessary, resulting in buffering of the excess data. However, most existing work on real time disk array access assume full striping[29] regardless of the individual file access requirements, sometimes in addition to high load[30] assumptions. The full striping assumption simplifies analysis by reducing the behavior of the array close to that of a single disk, but at the expense of using additional buffer and bandwidth. We are ....

P. S. Yu, M.-S. Chen, and D. D. Kandlur, "Grouped Sweeping Scheduling for DASD-based Multimedia Storage Management," Multimedia Systems, vol. 1, no. 3, 1993.

....requirement are also admitted in a batch. This approach provides better system throughput since only one media stream is needed to serve a batch of requests. We note that the admission criterion depends greatly on the disk scheduling policy used by the server (e.g. Grouped Sweeping Scheme [16], SCAN EDF [12] Greedy disk scheduling [13] However, since the purpose of this paper is not to show which admission control algorithms are the better ones, we implement a very simple admission criterion as follows: Estimated Service time Estimated Round duration Estimated Service time = c ....

P. S. Yu, M. Chen, and D. D. Kandlur. Grouped sweeping scheduling for dasd-based multimedia storage management. Multimedia Systems, 1:99--109, 1993.

....Australia, April 1 4, 1997. set of centralized servers, where a large volume of videos are stored. To ensure the jitter free service, the retrieval of videos from the storage subsystem and the delivery of them through the networks are carefully designed to meet their playback criterion [16, 17]. As a result, a video stream, corresponding to an I O stream and an isochronous channel, is typically reserved for a VOD service. Dedicating a stream for each viewer, however, will quickly exhaust the network I O bandwidth of the video server. This problem which limits the number of users a VOD ....

P. S. Yu, M. Chen and D. D. Kandlur. Grouped sweeping scheduling for DASD-based multimedia storage management. Multimedia Systems, 1:99--109, 1993.

....degree is necessary to stripe their blocks across all disks in the stripe group. Disk Scheduling The modeling results of [19] indicate that disk utilization over 80 can be achieved with essentially no stream starvation using EDF scheduling. A number of algorithms have been proposed ( 21] [22], 23] that provide higher disk throughput by grouping read requests for a number of streams, and scheduling each group to maximize throughput using a conventional scan algorithm. Modeling results presented in [24] indicate that for the large block sizes used by Tiger Shark (256KB and up) the ....

P. S. Yu, M. S. Chen, and D. D. Kandlur, Grouped sweeping scheduling for DASD-based multimedia storage management, in ACM Multimedia Systems Journal, 1:99-109, 1993.

....Under round robin scheduling, on the other hand, a stream can begin playback immediately upon receiving a block, since the arrival of its next block is exactly one round away. However, round robin may incur high seek times under random block placement. These observations lead Yu, Chen and Kandlur [81] to propose Grouped Sweeping Scheduling (GSS) In GSS, the N streams are divided statically into g groups. 33 Groups are served in round robin order, and SCAN is used within a group. Under GSS, a stream must wait only until all blocks in the group have been retrieved, suggesting that buffering ....

Philip S. Yu, Mon-Song Chen, and Dilip D. Kandlur. Grouped sweeping scheduling for DASD-based multimedia storage management. Multimedia Systems Journal, 1:99--109, 1993.

....by diskscheduling algorithms. Traditional disk scheduling algorithms such as rstcome rst serve and SCAN [15] 61] do not provide realtime guarantees. Hence, many disk scheduling algorithms have been proposed to address this issue. These include SCAN EDF [48] grouped sweeping scheduling (GSS) [77], and dynamic circular SCAN (DC SCAN) 30] which are described as follows. # The SCAN EDF combines the seek optimization of the traditional disk scheduling method SCAN [15] and the real time guarantees of the EDF mechanism. Note that the EDF mechanism in disk scheduling is non preemptive, which ....

P.S.Yu, M. S. Chen, and D. D. Kandlur, \Grouped sweeping scheduling for DASD-based multimedia storage management," ACM/Springer Multimedia Systems, vol. 1, no. 3, pp. 99-109, 1993.

....dt Twl ) minfx i T display g (15) ms X 1 (p i Gamma 1) x i S vf R dt Twl ] minf(p i Gamma k i ) x i T display g (16) By doing this, the retrieval of multiple streams can still be guaranteed. The recent proposed multimedia disk scheduling mechanisms by Chen, Kandlur and Yu in [22, 4, 21] targeted a solution to support independent concurrent media streams. By dividing the n requests into g groups, and serving groups in FIFO order and SCAN order within each group, this disk scheduling maintain the minimal disk seek overhead. Although the locations of different media streams can be ....

P. Yu, M. Chen, and D. Kandlur. Grouped sweeping scheduling for dasd-based multimedia storage management. ACM Multimedia Systems, 1(3):99--109, 1993. 31

.... rather than deterministic guarantees) Providing deterministic service at the disk is complicated by the random service time costs involved in disk transfers (because of the random seek and latency overheads) This problem has been addressed effectively by suitable disk scheduling policies [1, 2, 3, 4, 5, 6]. These scheduling 2 policies group a number of requests into rounds or batches and service the requests in a round using a disk seek optimizing policy such as SCAN. Then the service time for the entire round can be bounded to provide guarantees. This strategy works well with CBR streams. An ....

P. S. Yu, M. S. Chen, and D. D. Kandlur. Grouped sweeping scheduling for dasd-based multimedia storage management. Multimedia Systems, 1:99--109, 1993.

....ffl Together with scheme Stretch, we propose a novel disk scheduling policy that reduces the initial latency to minimum, regardless the number of requests in the system. This latency reduction technique does not consume additional system resources. ffl We revisit the Group Sweeping Scheme (GSS) [15], and after carefully analyzing its memory use, we show that it benefits greatly by also spacing out its IOs and by having large number of groups. This contradicts earlier results which indicated that a single group was best for GSS under high load situations. ffl We also study how so called ....

....TR ) Roughly half of the first factor, N Limit Theta S, is for the main buffer requirements; the rest is the requirement of the shared cushion buffers. Please refer to Appendix A.3 for the proof. 13 5. 2 Group Sweeping Scheme (GSS ) The Group Sweeping Scheme (GSS) proposed in [15] divides N Limit streams into G groups, with N Limit =G streams serviced in each group by a disk sweep. The groups are serviced in a round robin fashion. A request is assigned to a single group from the start to the end of its playback. In the published descriptions of GSS it is not clear to us ....

[Article contains additional citation context not shown here]

P. Yu, M.-S. Chen, and D. Kandlur. Grouped sweeping scheduling for DASD-based multimedia storage managemen. Multimedia Systems, 1(1):99--109, January 1993.

.... system of a Video On Demand (VOD) server needs to support the retrieval of video data to satisfy the continuous playback requirement (i.e. jitter free playback) Many issues have been investigated including disk scheduling schemes and multiple disks architecture to support jitter free playback [8, 5, 12, 14]. Most of the previous work assumed either a contiguous placement scheme or no data placement scheme (i.e. randomly placed) In this paper, we study the impact of the placement scheme in terms of the buffer size and number of users that can be supported. A mass storage system of a VOD server may ....

....the impact of the cluster (with the number of clusters equal to or greater than two) placement in terms of the number of users supported and the required buffer size. The analysis is based on NTSC quality video and an array of eight disks under FIFO, C SCAN, and GSS (Grouped Sweeping Scheduling) [14] disk scheduling schemes. The results showed that: 1. In FIFO scheme, the cluster placement can reduce the required buffer size up to 40 when compared to the contiguous placement. 2 2. In C SCAN scheme, the cluster placement can reduce the required buffer size up to 8 when compared to the ....

[Article contains additional citation context not shown here]

Yu P.S., Chen M.S., Kandlur D.D., "Grouped Sweeping Scheduling for DASD-based Multimedia Storage Management", Multimedia Systems, vol. 1, no. 3, 1993, pp 99-109. 29

....video streams, by empirically determining the number of clients that can be supported for different QoS and server parameters. There have been a number of different approaches for storage and retrieval of real time video data. Disk head scheduling and single disk data placement were addressed in [6, 8]. For the SPIFFI video server, 3] proposed a real time priority based disk scheduling algorithm that partitions requests into priority classes based on the nearness of their deadlines. In [1, 2] admission and resource allocation algorithms for VBR CDL video streams were studied in a singledisk ....

P. S. Yu, M-S. Chen, and D. D. Kandlur. Grouped Sweeping Scheduling for DASD-based Multimedia Storage Management. Multimedia Systems, 1(1):99--109, 1993.

.... were analyzed in [40, 41] Non contiguous disk allocation of streams was proposed and admission control based on constrained layout designs were analyzed in [37] A disk head scheduling algorithm, group sweeping scheme with the objective of minimizing buffer space and access time was presented in [47]. For the SPIFFI video server, 21] proposed a real time priority based disk scheduling algorithm that partitions requests into priority classes based on the nearness of their deadlines. These algorithms deal with scheduling a single disk. For multiple disk parallel disk arrays, simple reading of ....

P. S. Yu, M-S. Chen, and D. D. Kandlur. Grouped sweeping scheduling for DASDbased multimedia storage management. Multimedia Systems, 1(1):99--109, January 1993.

....starvation of certain requests. SCAN reduces the response time variations and optimizes seek times by serving the requests in an elevator like way. There exist many variations and hybrid solutions of the SSTF and SCAN algorithms that are widely used today [Geist et al. 87] Coffman et al. 90] Yu et al. 93] Modern disk scheduling algorithms [Jacobson et al. 91] Seltzer et al. 90] try to minimize the sum of seek and rotational delays by prioritizing, e.g. the request with the Smallest Positioning Time First. However, disk scheduling algorithms for multimedia data requests need to optimize, ....

Yu, P.S., Chen, M.S., Kandlur, D.D.: Grouped Sweeping Scheduling for DASD-Based Multimedia Storage Management, ACM Multimedia Systems, Vol. 1, No. 3, 1993, pp. 99-109

....very regular IO pattern, and this compensates for the poor seek overhead. In Chang and Garcia Molina, 1997a; Chang and Garcia Molina, 1997b, we analyze the memory requirement of Fixed Stretch and compare its performance with the performance of other disk scheduling policies (e.g. elevator and GSS Yu et al. 1993). We show that Fixed Stretch achieves throughput comparable to that of the other schemes, even though it has longer seek overhead. However, Fixed Stretch conserves memory, which is the critical resource, and this compensates for the high seek overhead. 3. 2 DIMENSIONAL BUBBLEUP (2DB) We now use ....

Yu, P., Chen, M.-S., and Kandlur, D. (1993). Grouped sweeping scheduling for DASD-based multimedia storage management. Multimedia Systems, 1(1):99--109.

....on some related work to support other aspects of VOD. Significant results were presented in [12] regarding admission control techniques and the choice of service size to support multi media applications. 13] examines the issue of collocational storage of multiple media segments on a disk. In [20], a new formulation for disk arm scheduling schemes called grouped sweeping scheduling is proposed and analyzed. The goal is to minimize the buffer requirement. A similar concept is also considered in [9] Furthermore, 16] studies storage management and disk access algorithms in a disk array ....

P. Yu, M.-S. Chen, D. Kandlur, "Grouped Sweeping Scheduling for DASD based Multimedia Storage Management", Multimedia Systems, vol. 1, no. 3, pp. 99-109, 1993.

....blocks and consequently have very poor resolution. In other related work, issues such as admission control and selection of service size to support multimedia applications are addressed in [5, 9] Disk scheduling schemes for optimizing the throughput of the DASD storage system are considered in [10, 11, 12]. This paper is organized as follows. In Section 2, the storage method and the playout method are presented. The segment sampling and segment placement schemes are devised in Section 3. Some experimental results are described in Section 4. This paper concludes with Section 5. 2 Video Data ....

....(i.e. the one starting from I frame 1) in the temporal order. Typically, the segment is large and comprises one or more disk tracks. It is advantageous to allocate entire disk tracks to segments since this results in a substantial increase in the access efficiency for storage and retrieval [3, 10]. Hence, the segment size is chosen such that it matches closely with a multiple of the disk track size. For example, given an MPEG stream of data rate 160 KBytes sec (approx. 1.5 Mbps) and a track size of 41 Kbytes, we choose a segment size of 15 frames which corresponds to 0.5 seconds of playout ....

P. S. Yu, M.-S. Chen, and D. D. Kandlur, "Grouped sweeping scheduling for DASD-based multimedia storage management," Multimedia Systems, vol. 1, pp. 99-- 109, 1993. Springer-Verlag.

No context found.

P.S. Yu, M.S. Chen and D.D. Kandlur. Grouped sweeping scheduling for DASD-based multimedia storage management. ACM Multimedia Systems, 1(3): 99-109, 1993.

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