John wiley sons reallyusefuls 0470857544 scalable continuous media streaming systems architecture design analysis and implementation jun 2005 ddu

Scalable Continuous Media Streaming Systems ARCHITECTURE, DESIGN, ANALYSIS AND IMPLEMENTATION Jack Y B Lee The Chinese University of Hong Kong, Hong Kong SAR, China Scalable Continuous Media Streaming Systems Scalable Continuous Media Streaming Systems ARCHITECTURE, DESIGN, ANALYSIS AND IMPLEMENTATION Jack Y B Lee The Chinese University of Hong Kong, Hong Kong SAR, China Copyright C 2005 John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex PO19 8SQ, England Telephone (+44) 1243 779777 Email (for orders and customer service enquiries): cs-books@wiley.co.uk Visit our Home Page on www.wiley.com All Rights Reserved No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, scanning or otherwise, except under the terms of the Copyright, Designs and Patents Act 1988 or under the terms of a licence issued by the Copyright Licensing Agency Ltd, 90 Tottenham Court Road, London W1T 4LP, UK, without the permission in writing of the Publisher Requests to the Publisher should be addressed to the Permissions Department, John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex PO19 8SQ, England, or emailed to permreq@wiley.co.uk, or faxed to (+44) 1243 770620 Designations used by companies to distinguish their products are often claimed as trademarks All brand names and product names used in this book are trade names, service marks, trademarks or registered trademarks of their respective owners The Publisher is not associated with any product or vendor mentioned in this book This publication is designed to provide accurate and authoritative information in regard to the subject matter covered It is sold on the understanding that the Publisher is not engaged in rendering professional services If professional advice or other expert assistance is required, the services of a competent professional should be sought Other Wiley Editorial Offices John Wiley & Sons Inc., 111 River Street, Hoboken, NJ 07030, USA Jossey-Bass, 989 Market Street, San Francisco, CA 94103-1741, USA Wiley-VCH Verlag GmbH, Boschstr 12, D-69469 Weinheim, Germany John Wiley & Sons Australia Ltd, 33 Park Road, Milton, Queensland 4064, Australia John Wiley & Sons (Asia) Pte Ltd, Clementi Loop #02-01, Jin Xing Distripark, Singapore 129809 John Wiley & Sons Canada Ltd, 22 Worcester Road, Etobicoke, Ontario, Canada M9W 1L1 Wiley also publishes its books in a variety of electronic formats Some content that appears in print may not be available in electronic books Library of Congress Cataloguing-in-Publication Data Lee, Jack Y B Scalable continuous media streaming systems : architecture, design, analysis and implementation / Jack Y.B Lee p cm Includes bibliographical references and index ISBN-13 978-0-470-85754-0 (HB) ISBN-10 0-470-85754-4 (HB) Streaming technology (Telecommunications) I Title TK5105.386.L44 2005 006.7 876–dc22 2005002759 British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library ISBN-13 978-0-470-85754-0 (HB) ISBN-10 0-470-85754-4 (HB) Typeset in 10/12pt Times by TechBooks, New Delhi, India Printed and bound in Great Britain by Antony Rowe Ltd, Chippeham, Wiltshire This book is printed on acid-free paper responsibly manufactured from sustainable forestry in which at least two trees are planted for each one used for paper production This book is dedicated to my wife Man-chu and our daughter Lok-sze for their love and support Contents Preface xvii Acknowledgements xix Part One: Fundamentals Introduction 1.1 Elements of a Multimedia System 1.2 Media Data 1.3 Media Delivery 1.4 Streaming versus Download 1.5 Challenges in Building Continuous Media Streaming Systems 1.5.1 Continuity 1.5.2 Known and Unknown Variations 1.5.3 Real-time Interactivity 1.5.4 Efficiency 1.5.5 Scalability 1.5.6 Reliability 1.6 Engineering Trade-offs 1.6.1 Trade-off in Capacity 1.6.2 Trade-off in Time 1.6.3 Trade-off in Space 1.6.4 Trade-off in Quality 1.6.5 Trade-off in Complexity 1.7 Performance Guarantee 1.8 Admission Control 1.9 Summary References 3 11 11 11 13 13 14 14 16 17 17 18 20 20 21 21 23 23 Media Compression 2.1 Introduction 2.1.1 Digital Audio 2.1.2 Digital Video 2.1.3 Media Compression 25 25 25 26 28 viii Contents 2.2 2.3 2.4 2.5 Media Multiplexing Temporal Dependencies in Compressed Video Bit-rate Variations Media Adaptation 2.5.1 Transcoding Techniques 2.5.2 Transcoder Design 2.5.3 Implementation Issues 2.5.4 Experimental Results 2.6 Summary References 28 30 31 33 33 34 36 37 40 40 Continuous Media Storage and Retrieval 3.1 Structure and Model of Hard Disk 3.2 Disk Scheduling 3.2.1 Performance Modeling 3.2.2 Capacity Dimensioning 3.3 Improving Disk Throughput 3.4 Grouped Sweeping Scheme 3.5 Multi-Disk Storage And Retrieval 3.5.1 Partition and Replication 3.5.2 Disk Striping 3.5.3 Multi-Disk Scheduling 3.6 Disk Zoning 3.7 Summary References 43 43 45 46 47 48 49 51 51 53 54 58 59 59 Soft Scheduling 4.1 Introduction 4.2 Statistical Capacity Dimensioning 4.3 Dual-Round Scheduling 4.3.1 Read-Ahead Algorithm 4.3.2 Performance Modeling 4.3.3 Buffer Requirement 4.4 Early-Admission Scheduling 4.4.1 Admission Algorithm 4.4.2 First-Block Replication 4.5 Overflow Management 4.5.1 Deadline-Driven Detection 4.5.2 Overflow Recovery 4.6 Performance Evaluation 4.6.1 Service Round Length Distribution 4.6.2 Statistical Streaming Capacity 4.6.3 Dual-Round Scheduling 4.6.4 Early-Admission Scheduling 4.6.5 Buffer Requirement 61 61 62 64 64 65 66 66 67 70 71 71 72 73 73 74 74 78 79 360 Scalable Continuous Media Streaming Systems L/4 Channel 1 Channel 91 92 Channel Channel … 30 2L/4 L 3L/4 31 32 33 … 60 61 62 63 … 90 91 92 93 … 120 93 … 120 31 61 62 63 … 90 61 62 63 … 90 91 92 93 … 120 31 32 33 … 60 61 62 63 … 90 91 92 … 30 32 33 … 60 … 30 93 … 120 31 32 33 … 60 … 30 Data retrieved in a service round B 1,j,k B 2,j,k …… b i.j + kN +(Nsc −1)H/Nsc B M,j,k Disk Disk Disk Disk N G1,0 G2,0 G3,0 GN,0 G1,1 G2,1 G3,1 G1,2 G2,2 G3,2 GN,2 G1,3 G2,3 G3,3 GN,3 …… G j,o …… …… b i j + kN + H/Nsc …… b i j +kN …… B i,j,k …… GN,1 Figure 20.3 Data access pattern and placement of static channels 20.4.2 Interleaving of Data Blocks Next we consider the interleaving data placement policy for serving the static channels Consider a system with N = 2, N S = and k = 120 as shown in Figure 20.3, illustrating the data access pattern for the static channels We observe that data blocks bi, j , bi, j + 30 ,bi, j + 60 , bi, j + 90 of video i stored in disk and data blocks bi, j+1 , bi, j + 31 , bi, j + 61 , bi, j + 91 stored in disk are always retrieved together in the same service round, with j ∈ [1, 3, 5, 29] Thus, by placing these data blocks in a continuous portion of the disk surface, we can effectively eliminate the disk seeks required in conventional round-based schedulers However, this placement policy only works for static channels where the transmission schedules are known and fixed We address data retrievals for dynamic channels using replication in the next section 20.4.3 First T R Seconds Replication Data retrievals for dynamic channels are more random in nature and hence the interleaving data placement policy offers no advantage Moreover, with the WSGP policy in place, serving the dynamic channels using the interleaving data placement policy will result in the disk constantly seeking between outer tracks and inner tracks, further degrading disk throughput To tackle this problem, we note that dynamic channels have one crucial property – it only serves up to Efficient Server Design for Hybrid Multicast Streaming 361 Zones used to accommodate first TR seconds of data (Zdyn =3) Zone Zone Zone Zone Zone Zone Zone Zone Zone Zone and from a zone pair in WSGP to accommodate data for static channels Zone and from a zone pair in WSGP to accommodate data for static channels Zone and from a zone pair in WSGP to accommodate data for static channels Figure 20.4 Data layout for a disk with Z = and Z dyn = the first TR seconds of a video Therefore, we propose replicating the first TR seconds of each video in the outermost zones of the disk, thereby taking advantage of the higher transfer rate of the outermost zones Assume the first Z dyn zones are used to store the replicated video data, then the WSGP algorithm will begin pairing zone (Z dyn + 1) and Z Figure 20.4 illustrates the overall data layout 20.4.4 An Integrated Scheduler To support data retrievals for both static and dynamic channels, we devise a new integrated scheduler based on the three design features previously discussed The integrated scheduler is still round-based but each round is divided into two parts – a static round and a dynamic round as shown in Figure 20.5 In a static round, two continuous data retrievals are performed, one for an outer-track block and one for an inner-track block The retrieved data will then be used for transmission over the static channels The dynamic round is further sub-divided into GD dynamic micro-rounds and the dynamic channels are then assigned to these GD dynamic micro-rounds as in the GSS case Retrievals within a dynamic micro-round will be executed using SCAN and the retrieved data will be transmitted over the dynamic channels The buffer requirement of this scheduler is thus given by NQ(2N S + N D ) as illustrated in Figure 20.6 (20.4) 362 Scalable Continuous Media Streaming Systems TR TR static channel static channel static channel Micro-round Time A macro-round Static round All static channels start transmit at the same time Transmission Static round Dynamic round Static round … Data Retrieval Data from outer zones Data from inner zones Data for all the requests in a groups Micro-round Figure 20.5 Scheduling disk retrieval and network transmission of WSGP Data in cache NQ(2NS+ND) Buffer requirement NQ(NS+ND) NQ(NS+2ND /GD) NQ(NS+ND /GD) NNSQ Dynamic round Dynamic round Static round Static round Time : Transmissions Figure 20.6 Data in cache at different temporal positions in a service round This scheduler, however, has a subtle problem We found that the server buffer requirement is dominated by the memory used to cache data for the static channels, which involved continuous retrievals for two large data blocks, and this increases the buffer requirement significantly To reduce the buffer requirement, we sub-divide the static round into G S micro-rounds of equal durations, where G S equals to the number of videos In each static micro-round, static channels belonging to the same video are scheduled and data are transmitted at the end of the static Efficient Server Design for Hybrid Multicast Streaming 363 NCQ(1+1/G) Data in cache NCQ 3NCQ/G Buffer requirement 2NCQ/G NCQ/G Time micro-round micro-round service round micro-round micro-round service round Transmissions Figure 20.7 Server buffer requirement when micro-rounds are of different durations (G = 4) micro-round Since different groups retrieve data from different zones of the disk, the static micro-round will be of different durations As illustrated in Figures 20.7 and 20.8, and using derivations similar to GSS we can derive the new total buffer requirement, which is reduced to NQ N S + GS + ND + GD (20.5) In practice, this modification can reduce the buffer requirement of static channels by as much as 40% In the next section we evaluate and compare performance of the proposed server design with the GSS-based server design 20.5 Performance Evaluation In this section, we evaluate the performance of the presented efficient server design and compare it against the GSS-based server design Table 20.1 lists the key system parameters used in the numerical calculations and Table 20.2 gives the specification of the disks used in performance evaluation Table 20.1 System parameters used in performance evaluation System parameter Video data rate Number of disk Length of video Static channel per video Dynamic channel per video Symbol Value RV N L NS ND 4Mb/s (MPEG2) 7200s 20 20 Table 20.2 Specification of different disks used in performance evaluation Disk Parameter Value Disk Model Atlas 10K Barracuda Cheetah 9LP IBM 18es Disk rotation speed (rpm) Full strobe seek time (ms) Track-to-track seek time (ms) Head switching time (ms) Number of data surfaces Blocks per disk Total number of tracks Number of zones 10,025 10.828 1.245 0.176 17,938,986 10,022 24 7,200 16.679 1.943 0.100 4,110,000 5,172 11 10.045 10.627 0.831 0.030 12 17,783,240 6,962 11 7200 12.742 1.086 0.062 17,916,240 11,474 55 Data in cache NCQ(1+1/G) NCQ 3NCQ/G 2NCQ/G NCQ/G Time micro-round micro-round service round service round Transmission Figure 20.8 Server buffer requirement when micro-rounds are of same durations (G = 4) Efficient Server Design for Hybrid Multicast Streaming 365 1000 800 Channel Supported 600 400 200 0 400 800 1200 Server Buffer (in MB) 1600 2000 Our Efficient Server Design (Atlas10k) Our Efficient Server Design (Barracuda) Our Efficient Server Design (Cheetah9LP) Our Efficient Server Design (IBM18es) GSS Only (Atlas10k) GSS Only (Barracuda) GSS Only (Cheetah9LP) GSS Only (IBM18es) Figure 20.9 Server capacity versus server buffer constraint for four hard disk models 20.5.1 Server Capacity Figure 20.9 plots the server capacity versus the server buffer size constraint for our efficient server design and the GSS-based server design for four different disks Compared to GSS-based design, our design can increase the server capacity by up to 60% Moreover, the performance gain increases to about 40% on average when the buffer available is more than 1GB as evident in Figure 20.10 This is because the interleaved data placement policy reduces disk seeks substantially and thus the gain in I/O efficiency due to larger block size becomes more significant 20.5.2 Utilization of Disk Capacity The major advantage of using WSGP is that while achieving full utilization of disk storage capacity, it provides similar performance improvement as SGP Results show that with the same system configuration, SGP can only utilize 91% storage capacity of the disk while WSGP can achieve 100% utilization 366 Scalable Continuous Media Streaming Systems Percentage Increase 70% 60% 50% 40% 30% 20% 10% 0% 400 600 800 1000 1200 1400 1600 1800 2000 Server Buffer (in MB) Atlas10k Cheetah 9LP Barracuda IBM18es Figure 20.10 Percentage increase in number of channels supported compared with GSS 20.6 Summary In this chapter, we presented an efficient disk-array-based server design for the Super-Scalar VoD system We proposed a placement scheme to exploit disk zoning and the characteristics of static and dynamic channels Coupled with an integrated scheduler, we were able to increase the server capacity by as much as 60% compared to the conventional GSS-based server design While the server design presented in this chapter is specifically targeted for use in a SS-VoD system, the design principles are general and thus can be applied to other multicast video streaming architectures with both periodic and aperiodic multicast streaming channels References [1] K Breidler, H Kosch, and L Băoszăormenyi, A Comparative Study of Selected Parallel Video Servers, Proceedings of IEEE 2000 11th International Workshop on Database and Expert Systems Applications, September 6–8, 2000, Greenwich, London [2] D.J Gemmell, H.M Vin, D.D Kandlur, P.V Rangan, and L.A Rowe, Multimedia Storage Servers: A Tutorial, IEEE Computer, vol 28, no 5, May 1995, pp 40–49 [3] W.J Bolosky, J.S Barrera III, R.P Draves, R.P Fitzgerald, G.A Gibson, M.B Jones, S.P Levi, N.P Myhrvold, and R.F Rashid, The Tiger Video Fileserver, Proceedings of Sixth International Workshop on Network and Operating System Support for Digital Audio and Video, IEEE Computer Society Press, Los Alamitos, CA, 1996 [4] Z.-R Lin and M.-S Chen, Design and Performance Study of Scalable Video Storage in a Disk-Array-Based Video Server, Proc of International Conference on Multimedia and Expo (ICME 2000), vol 3, 2000, pp 1341– 1344 Efficient Server Design for Hybrid Multicast Streaming 367 [5] J.B Kwon and H.Y Yeom, Generalized Data Placement for Periodic Broadcast of Videos, Proc IEEE International Conference on Multimedia and Expo (ICME 2001), August, 2001, Tokyo, Japan [6] P.S Yu, M.S Chen and D.D Kandlur, Grouped Sweeping Scheduling for DASD-Based Multimedia Storage Management, ACM Multimedia Systems, vol 1, no 3, 1993, pp 99–109 [7] N Reddy and A.L Wyllie, ‘I/O Issues in a Multimedia System,’ IEEE Computer, vol 27, no 3, March 1994, pp 69–74 [8] S Chen and T Manu, A Novel Video Layout Strategy for Near Video-on-Demand Servers, Proceedings of IEEE International Conference on Multimedia Computing and Systems’ 97, Ottawa, Canada, June 1997, pp 37–45 [9] S.L Tsao and Y.M Huang, An Efficient Storage Server in Near Video-on-Demand Systems, IEEE Transactions on Consumer Electronics, vol 44, no 1, Feb 1998, pp [10] G.A Gibson, Redundant Disk Arrays: Reliable, Parallel Secondary Storage, MIT Press, 1992 Index Active Disk Synchronization (ADS), 101–102 Admission complexity, 124 Admission control, 21, 132, 333 observational, 23 Admission test, 132 Admission threshold, 334, 344 Admission-scheduler-based (ASB) protocol, 197 Aggregated Monotonic Decreasing Rate Scheduler (AMDR), 130 Asynchronous Grouped Sweeping Scheme (AGSS), 180 See also disk scheduling algorithm ATM (Asynchronous Transfer Mode), 164, 174, 176, 196 B frame(s), 30 Bandwidth overhead, 240 Bandwidth partition scheme, 310 consonant broadcasting, 315 Baseline rebuild, 252 Batching, 295 Best-effort service, 21 Bit-rate conformance, 39 Bit-rate variations due to compression, 31 playback, 305 smoothing of, 119 Broadcast schedule, 310 Broadcasting, 287 channel, 288 consonant, 313 medium, 288 periodic, see open-loop algorithms server design for, 356 service, 288, 291 TV, 288 Buffer constraint, 120 Buffer sharing, 95 Caching, 303 in consonant broadcasting, 315 in multicast streaming, 292, 298–299 in redundant data update, 278 in SS-VoD, 337 prefix, 304 Capacity dimensioning, 47, 91 effect of disk zoning on, 58 statistical, 62 Channel partitioning, 344, 346 Channel requirement, 348 Channel switching latency, 327 Client pull, 167 See also service model Client-server model, Client-server ratio, 174 Clock jitter, 175, 219 Closed-loop algorithms, 293, 295 See also multicast streaming Common Interchange Format (CIF), 27 Compression, 25 audio, 25 complexity, 20 efficiency, 20 constant-bit-rate, 32 constant-quality, 32, 165 Scalable Continuous Media Streaming Systems Jack Y B Lee C 2005 John Wiley & Sons, Ltd 370 Compression (Continued ) inter-frame, 30 intra-frame, 30 layered video coding, 33, 141 lossless, 26 lossy, 26 standards, 25, 28 variable-bit-rate, 32, 165 Concurrent push architecture, 173 compare to staggered push, 226–230 Concurrent schedule, 54 See also multi-disk scheduling Congestion control RAP, 142 TCP, 112 Consonant Broadcasting (CB), 313 See also multicast streaming Constant-bit-rate encoding, 32 See also compression Constant-quality encoding, 32, 165 See also compression Continuity condition, 47 playback, 10–11 Continuous media, C-SCAN, 46 See also disk scheduling algorithm Cumulative data consumption function, 120 Data partition scheme, 310 Data rebuild for disk, 92 for server, 249 Data reorganization, 263, 265 overhead, 270 Data units (blocks), see erasure(s) Deadline-Driven Detection, 71 Decoding time deviation, 177 peak-to-peak, 178 Degraded mode operation, 84 Delay budget, Detection delay, 196 effect on buffer requirement, 208 modeling of, 240-241 Deterministic performance guarantee, 21 See also QoS guarantee Index Deviation bound, disk asynchrony, 100 Direct Streaming Transfer (DST), 26 Discrete media, See also continuous media Disk arm, 43 controller, 44 head, 43 platter, 43 sector(s), 43 track(s), 43 Disk migration, 252 Disk model, 43, 90 parameters, 103, 364 read-on-arrival, 92 rotational latency, 44 seek time, 44 transfer rate, 44 worst-case seek time, 47, 62, 91 zoning, 58, 61 Disk scheduling algorithm asynchronous grouped sweeping scheme (AGSS), 180 C-SCAN, 46 first-come-first-serve (FCFS), 45 grouped sweeping scheme (GSS), 49 integrated scheduler for hybrid server, 361 Distributed rebuild, 254 Distributed sparing, 250 See also sparing scheme(s) Download model, Drift compensation, 36 Dual-round scheduling (DRS), 64 DVD Audio, 25 Dynamic multicast channel, 332 Dynamically admitted, 334 waiting time for, 341 Early-Admission Scheduling (EAS), 66 non-preemptive schedule, 69 preemptive schedule, 69 Elastic traffic, Elevator seeking, 46 See also disk scheduling algorithm Index Erasure(s), 194 correction process, 88 Reed-Solomon Erasure Correction (RSE) code, 275 Erlang-k distribution, 242 Excess redundancies, 240 Fail-stop, 194 Failure-detection protocol, 197, 230 admission-scheduler-based (ASB) protocol, 197 detection delay, 196 Filling time, average, 176 First-Block Replication (FBR), 70 First-come-first-serve (FCFS), 45 See also disk scheduling algorithm Forward erasure correction (FEC), 169, 195, 238 Group of pictures (GOP), 31 Grouped consonant broadcasting (GCB), 320 Grouped Sweeping Scheme (GSS), 49 See also disk scheduling algorithm H.264, 28 See also compression, standards Hard scheduling, 61 High-Definition Video, 28 Hot sparing, 250 See also sparing scheme(s) Hyper-Text Transfer Protocol (HTTP), 112 I frame(s), 30 IGMP join group, 290, 327 leave group, 327, 333 Independent proxy, 162 Inelastic traffic, Interactive control, 339 pause-resume, 339 seeking, 340 slow motion, 339 Interactive multicast streaming, 287 Inter-frame compression, 30 See also compression Internet phone, Internet2, 289 371 Intra-frame compression, 30 See also compression IP multicast, 290 Layered-video codec, 33, 141 See also compression Lossless smoothing algorithm, 121 Macro-round, 50, 218, 357 Maximum advance, 178 Maximum lag, 178 Maximum Queue Length (MQL), 297 MDR transmission schedule, 123 Media adaptation, 33 Media delivery real-time, 5, 13 soft-real-time, 5–6 Media multiplexing, 28 Meridian Lossless Packing (MLP), 26 Micro-round, 50, 199, 218, 357 Microsoft Media Services (MMS), 113 See also streaming protocol(s) Min-Rate transmission, 196 Mixed-distributed-baseline rebuild, 255 Monotonic Decreasing Rate Scheduler (MDR), 122 MPEG multiplexer, 29 MPEG-1, 28 MPEG-2, 28 MPEG-3, 28 MPEG-4, 28 MTTF (mean time to failure), 237, 239 MTTR (mean time to repair), 239 Multicast address (group), 290 Multicast routing protocols, 290 Multicast streaming broadcasting, 287 interactive multicast streaming, 287 on-demand multicast streaming, 287 Multi-disk scheduling concurrent schedule, 54 offset schedule, 56 split schedule, 56 Multi-row-permutated data reorganization (m-RPDR), 269 Multistream pipelining, 10 372 Non-stop service despite disk failure, 89 despite server failure, 193 Normal mode operation, 84 Normalized capacity gain, 74 Normalized latency, 346 NTSC, 27 NVoD (Near-Video-on-Demand), 331–332 Observational admission control, see admission control Offset schedule, 56 See also multi-disk schdeuling On-demand multicast streaming, 287 See also multicast streaming Open-loop algorithm(s), 309 See also multicast streaming Overflow probability constraint, 62 Overflow recovery, 72 Over-rate transmission (ORT), 222 P frame(s), 30 Parity group, 85, 238, 273, 275 reshuffling, 276 Partition, data, 51 Patching, 297 See also multicast streaming recursive, 301 transition, 300 stream (P-stream), 300 Pause-resume, 339 See also interactive control Performance guarantee best effort, 21 deterministic, 21 probabilistic (statistical), 21 Periodic broadcasting, 309 See also multicast streaming Piggybacking, 304 See also multicast streaming Pipelined rebuild, 99 See also data rebuild Placement policy randomized, 62, 166 round-robin, 165, 166, 174, 218, 265 row-permutated, 265, 269 weighted segment group pairing (WSGP) scheme, 359 Index Playback buffers, 94 Prefetch, 113, 144, 150, 178 See also prefill delay Prefill delay, 180, 185, 226 See also prefetch Prefix, 304 caching, 304 See also caching proxy, 304 Probabilistic performance guarantee, 21 See also performance guarantee Progressive redundancy transmission (PRT), 169, 196, 203, 240 Proxy-at-client, 163 Proxy-at-server, 161 QoS guarantee, 13, 141, 176 See also performance guarantee Quantizer regulation, 37 Quarter-Common Interchange Format (QCIF), 27 Read-on-arrival, 92 See also disk model RealNetworks Data Transport (RDT), 114 See also streaming protocol(s) Real-time delivery, 5, 13 Internet phone, video conferencing, Real-time streaming protocol (RTSP), 114 See also streaming protocol(s) Real-time transport protocol (RTP), 115 See also streaming protocol(s) Rebuffering ratio, 149 Rebuild buffers, 94 rate, 94 time, 93 Rebuild algorithms baseline rebuild, 252 distributed rebuild, 254 mixed-distributed-baseline rebuild, 255 Rebuild mode operation, 84 Reception schedule, 310 Recursive patching, 301 See also multicast streaming Redundant Array of Inexpensive Disks (RAID), 86 Index Redundant data update, 272 for multiple redundant nodes, 280 overhead, 278 Redundant server scheme (RSS), 204 Redundant units (blocks), see erasure(s) Reed-Solomon Erasure Correction (RSE) code, 275 See also erasure(s) Regular stream (R-stream), 300 Reliability challenge, 14 parallel server architecture, 193 storage systems, 83 Replication, 51 Requantization threshold bit-rate (RTB), 35 Resource allocation, 23, 131, 196, 230 Resource reservation, see resource allocation Response time, system, 180, 226 Rotational latency, 44 See also disk model Round-robin placement, 165, 166, 174, 218, 265 See also placement policy Row-permutated data reorganization (RPDR), 265, 269 See also placement policy RTP control protocol (RTCP), 115 See also streaming protocol(s) SCADDAR, 264 See also placement policy Scalability, 14 Schedulig delay, 180 Seek disk seek, 44 interactive control, 340 Server failure, see fail-stop Server push, 167, 174 See also service model Server rebuild, 249 See also data rebuild Service group, 357 Service model, 174 client pull, 167, 174 server push, 167, 174 Short striping, 164 See also striping Skewness, video popularity, 174 See also Zipf Slow motion, 339 See also interactive control Smoothing, of video bit-rate, 119 373 Soft scheduling, 61 See also disk scheduling algorithm Soft-real-time delivery, 5–6 video-on-demand, Sparing scheme(s), 93, 250 distributed sparing, 250 hot sparing, 250 Split schedule, 56 See also multi-disk scheduling SS-VoD (Super-Scalar Video-on-Demand), 331 See also multicast streaming Staggered push, 217 Start-up latency, 302, 318, 340 Static multicast channel, 332 Statically admitted, 334 waiting time for, 340 Statistical capacity dimensioning, 62 See also capacity dimensioning Statistical performance guarantee, 21 Std-Rate transmission, 196 Streaming in mixed-traffic networks, 121 model, streaming protocol(s) Microsoft media services (MMS), 113 RealNetworks data transport (RDT), 114 Real-time streaming protocol (RTSP), 114 Real-time transport protocol (RTP), 115 Striping disk striping, 53 network striping, 164 server striping, 164 short striping, 164 space striping, 165 sub-frame striping, 165 sub-schedule striping, 184 tape striping, 164 time striping, 164 wide striping, 164 sub-schedule striping (SSS), 184 Super Audio CD (SACD), 26 Synchronization active disk synchronization (ADS), 101–102 media streams, server, 168 374 System expansion, 263 data reorganization, 265 redundant data update, 272 System reconfiguration, 196 System restoration, 250 Taxonomy open-loop multicast streaming algorithms, 310 parallel server architecture, 159 Temporal dependencies, in compressed video, 30 Test Model (TM5), 37 Time-scale modification, 305 Track group, 100 Track-based rebuild, 96 Track-based retrieval, 48 Trade-off, 16 capacity, 17, 51, 62, 269, 324 complexity, 20-21, 296, 350 quality, 20, 33, 304 space, 18, 48, 51, 66, 122, 237, 299 time, 17, 48, 295 Traffic overlapping, 222 Transcoder, 33 design, 34 implementation issues, 36 Transcoding techniques, 33 requantization, 34 spatial downscaling, 34 Transition patching, 300 Index Transition stream (T-stream), 301 Transmission control protocol (TCP), 111 congestion control, 112 Transmission jitter, 176 Transmission overhead, of FEC, 169, 240 Transmission schedule, 120, 124 TV broadcasting, 288 See also broadcasting TVoD (True-Video-on-Demand), 291 Uneven group assignment, 181 Unicast, 287 User Datagram Protocol (UDP), 113 UVoD (Unified Video-on-Demand), 344 Vandermonde matrix, 275 variable-bit-rate encoding, 32, 165 See also compression Video composition, 26 standards, 27 block consumption model, 177, 224 conferencing, placement problem, 52 Weighted segment group pairing (WSGP) scheme, 359 Wide striping, 164 See also striping Zipf, 52 Zoned bit recording (ZBR), 58, 61 ... first one, called DVD audio [38], is Scalable Continuous Media Streaming Systems Jack Y B Lee C 2005 John Wiley & Sons, Ltd 26 Scalable Continuous Media Streaming Systems Figure 2.1 A sequence of.. .Scalable Continuous Media Streaming Systems ARCHITECTURE, DESIGN, ANALYSIS AND IMPLEMENTATION Jack Y B Lee The Chinese University of Hong Kong, Hong Kong SAR, China Scalable Continuous Media. .. Preface This book addresses the architecture, design, analysis, and implementation of scalable and reliable continuous media streaming systems This is an intermediate to advanced book aimed at