Click here to load reader

Analytic Evaluation of Quality of Service for On-Demand Data Delivery Hongfei Guo ([email protected]) Haonan Tan ([email protected])

  • View
    213

  • Download
    0

Embed Size (px)

Text of Analytic Evaluation of Quality of Service for On-Demand Data Delivery Hongfei Guo ([email protected])...

  • Analytic Evaluation of Quality of Service for On-Demand Data Delivery Hongfei Guo ([email protected])Haonan Tan ([email protected])

    CS747 Project Presentation

  • OutlineBackgroundTwo Multicast ProtocolsCustomized MVA AnalysisValidationModel Improvement (Interpolation) Evaluation of Different Multicast Protocols Conclusion & Future Work

    CS747 Project Presentation

  • BackgroundEager et al. reasoned minimum bandwidth requirements. But How about Quality of Service ? Balking probability Waiting timeGiven: server bandwidth multicast protocol

    CS747 Project Presentation

  • Two Multicast ProtocolsGrace Patching Shared multicast stream (current data) Unicast patch stream (missed data) Average required server bandwidth

    CS747 Project Presentation

  • Two Multicast Protocols (contd)Hierarchical Multicast Stream Merging Each data transmission stream is multicast Clients accumulate data faster than file play rate Clients merged into larger and larger groups Once merged, clients listen to the same streams Average required server bandwidth

    CS747 Project Presentation

  • CMVA AnalysisCustomer Balking ModelFixed number of streams in the serverAn arriving customer leaves if no streams availableCustomer Waiting ModelFixed number of streams in the serverAn arriving customer waits till it being servedCustomers with same request coalesce in the waiting queue

    CS747 Project Presentation

  • Input ParametersC server capacity external customer arrival rateMnumber of file categories For i = 1, 2, , MKithe total number of distinct files in category iTimean duration of the entire file in category iizipfian parameter in category iPiprobability accessing category i files

    CS747 Project Presentation

  • Output Parameters (Balking)

    S1average service time at center 1R0mean residence time at center 0X system throughput. For i = 1, 2, #files on the server pifraction of customer requests for file iCiaverage b/w for file iS1imean service time of file i streams at center 1S0mean service time at center 0Q0mean queue length at center 0Xi throughput of streams serving file iPBmean incoming costumer balking probability

    CS747 Project Presentation

  • Output Parameters (Waiting)Wmean waiting time for a request (not coalesced)U system utilizationSoverall mean stream duration estimateFor i = 1, 2, , #files on the server pifraction of customer requests for file iSimean stream duration for file iQimean number of waiting requests (not coalesced) for file iXimean throughput of requests (not coalesced) for file iRimean residence time of a request (not coalesced) for file iCiaverage number of active streams for file iRimean residence time adjusted for coalescingWimean waiting time adjusted for coalescing

    CS747 Project Presentation

  • (1) Customer Balking ModelCenter 0 SSFR center Represent the waiting state of a streamCenter 1 Delay center Represent the active state of a stream

    CS747 Project Presentation

  • CMVA Equations (Protocol result)(interarrival time)

    CS747 Project Presentation

  • (2) Waiting ModelCenter 0 multi-channel server with C streamsTwo kinds of measurements (from two perspectives)Server only see non-coalesced customer requestsCustomers count in both coalesced and non-coalesced requests. Center 0

    CS747 Project Presentation

  • CMVA EquationsMeasurements for the server

    CS747 Project Presentation

  • CMVA Equations (contd)Measurements for the customers

    CS747 Project Presentation

  • Validation (1)

    CS747 Project Presentation

    Chart1

    0.9760.9762525

    0.9310.9175050

    0.8670.7877575

    0.7840.5161000.2

    0.6820.1161250.039

    0.5610.0281500.001

    0.4230.0131750

    0.2720.0082000

    0.1240.0060.20

    0.0430.0030.0160

    0.020.0020.0010

    0.0120.00200

    Patching

    HMSM(2,1)

    Sim-P

    Sim-H

    Capacity

    Probability

    Balking Probabilities

    balking-eval

    PatchingBalking

    lamda=2000T=1

    c'234.241307.455396.224499.719

    204080160Avg. Total B/W

    0.02950.9159

    0.05850.717117

    0.1030.234206

    0.11750.080.08235

    0.1320.027264

    0.2340.002468

    0.46850937

    0.03850.90177

    0.0770.701154

    0.1350.226270

    0.1540.0710.071308

    0.1730.021346

    0.30750.001615

    0.61501230

    0.04950.88999

    0.0990.684198

    0.17370.218347

    0.19850.0610.061397

    0.22330.017446

    0.3960.001792

    0.792501585

    0.06250.874125

    0.1250.662250

    0.2190.204438

    0.250.0550.055500

    0.280.014560

    0.49950.001999

    0.999501999

    SkimmingBalking

    c'121.55192.787290.747412.337

    204080160

    0.0150.96730

    0.03050.87261

    0.05340.42106

    0.0610.15122

    0.0690.047138

    0.12150.003243

    0.2430486

    0.0240.94948

    0.0480.82896

    0.0840.36168

    0.09650.115193

    0.10860.033217

    0.1930.002386

    0.38550771

    0.03650.92673

    0.07250.776145

    0.1270.301254

    0.14550.086291

    0.1820.011364

    0.29050.001581

    0.581501163

    0.05150.902103

    0.1030.722206

    0.18070.252361

    0.20650.066413

    0.23230.017464

    0.41250.001825

    0.824501649

    PatchingBalkingUtil

    204080160Avg. Total B/W

    0.02950.998

    0.05850.996

    0.1030.984

    0.11750.9530.953

    0.1320.874

    0.2340.5

    0.46850.25

    0.03850.998

    0.0770.997

    0.1350.987

    0.1540.9580.958

    0.1730.878

    0.30750.499

    0.6150.25

    0.04950.998

    0.0990.997

    0.17370.989

    0.19850.9620.962

    0.22330.88

    0.3960.5

    0.79250.25

    0.06250.999

    0.1250.998

    0.2190.991

    0.250.9650.965

    0.280.884

    0.49950.5

    0.99950.25

    SkimmingBalkingutil

    c'121.55192.787290.747412.337

    204080160

    0.0150.998

    0.03050.996

    0.05340.987

    0.0610.953

    0.0690.87

    0.12150.5

    0.2430.25

    0.0240.998

    0.0480.997

    0.0840.989

    0.09650.958

    0.10860.879

    0.1930.499

    0.38550.25

    0.03650.998

    0.07250.997

    0.1270.99

    0.14550.962

    0.1820.795

    0.29050.5

    0.58150.25

    0.05150.999

    0.1030.998

    0.18070.991

    0.20650.965

    0.23230.882

    0.41250.5

    0.82450.25

    balking-eval

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    20 files

    40 files

    80 files

    160 files

    Avg. Total B/W

    Available Server Bandwidth per Client

    Balking Probability

    Patching Balking

    waiting-eval

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    20

    40

    80

    160

    Available Server Bandwidth per Client

    Balking Probability

    Skimming Balking

    balking-patching

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    00000

    20 files

    40 files

    80 files

    160 files

    Avg. Total B/W

    Available Server Bandwidth per Client

    Server Utilization

    Patching Balking

    sim-patching

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    0000

    20

    40

    80

    160

    Available Server Bandwidth per Client

    Server Utilization

    Skimming Balking

    PatchingWaitingTimePatchingWaitingUtil

    lamda=2000T=1lamda=2000T=1

    c'234.164307.269396.208499.672c'

    204080160204080160

    0.02950.0640.02950.998

    0.05850.0160.05850.997

    0.11700.1170.987

    0.117500.11750.986

    0.23400.2340.5

    0.468500.46850.25

    0.03850.120.03850.999

    0.0770.0310.0770.998

    0.153500.15350.99

    0.15400.1540.989

    0.307500.30750.5

    0.61500.6150.25

    0.04950.2180.04950.999

    0.0990.0580.0990.998

    0.19800.1980.992

    0.198500.19850.991

    0.39600.3960.5

    0.792500.79250.25

    0.06250.3850.06250.999

    0.1250.1040.1250.998

    0.249500.24950.993

    0.2500.250.992

    0.499500.49950.5

    0.999500.99950.25

    SkimmingWaitingTimeSkimmingWaitingUtil

    c'121.55192.787290.747412.337c'121.55192.787290.747412.337

    204080160204080160

    0.0150.1980.0150.999

    0.03050.0510.03050.998

    0.06050.0010.06050.985

    0.0610.0010.0610.983

    0.121500.12150.5

    0.24300.2430.25

    0.0240.2690.0240.999

    0.0480.0780.0480.998

    0.0960.0010.0960.989

    0.09650.0010.09650.988

    0.19300.1930.499

    0.385500.38550.25

    0.03650.3770.03650.999

    0.07250.1170.07250.999

    0.1450.0010.1450.992

    0.14550.0010.14550.991

    0.290500.29050.5

    0.581500.58150.25

    0.05150.5580.05150.999

    0.1030.1770.1030.999

    0.2060.0010.2060.994

    0.20650.

Search related