26
Parallelizing Video Transcoding With Load Balancing On Cloud Computing Song Lin, Xinfeng Zhang, Qin Y, Siwei Ma Circuits and Systems, 2013 IEEE

Parallelizing Video Transcoding With Load Balancing On Cloud Computing Song Lin, Xinfeng Zhang, Qin Y, Siwei Ma Circuits and Systems, 2013 IEEE

Embed Size (px)

Citation preview

Page 1: Parallelizing Video Transcoding With Load Balancing On Cloud Computing Song Lin, Xinfeng Zhang, Qin Y, Siwei Ma Circuits and Systems, 2013 IEEE

Parallelizing Video Transcoding With Load Balancing On Cloud Computing

Song Lin, Xinfeng Zhang, Qin Y, Siwei Ma

Circuits and Systems, 2013 IEEE

Page 2: Parallelizing Video Transcoding With Load Balancing On Cloud Computing Song Lin, Xinfeng Zhang, Qin Y, Siwei Ma Circuits and Systems, 2013 IEEE

Outline

Introduction

Related work

Problem formulation and system architecture

Proposed method

Experiment Results

Conclusion

Page 3: Parallelizing Video Transcoding With Load Balancing On Cloud Computing Song Lin, Xinfeng Zhang, Qin Y, Siwei Ma Circuits and Systems, 2013 IEEE

Introduction #1

Parallel programmingShare memory

Pthread – data dependency

Message passing MPI – time delay

Page 4: Parallelizing Video Transcoding With Load Balancing On Cloud Computing Song Lin, Xinfeng Zhang, Qin Y, Siwei Ma Circuits and Systems, 2013 IEEE

Introduction #2

IssuesData dependencyCost of data passingLoad balance

Page 5: Parallelizing Video Transcoding With Load Balancing On Cloud Computing Song Lin, Xinfeng Zhang, Qin Y, Siwei Ma Circuits and Systems, 2013 IEEE

Introduction #3

Cloud computationData segmentationComputing capacity

Page 6: Parallelizing Video Transcoding With Load Balancing On Cloud Computing Song Lin, Xinfeng Zhang, Qin Y, Siwei Ma Circuits and Systems, 2013 IEEE

Introduction #4

GOP-based encodingIndependence between GOPs

...........

Page 7: Parallelizing Video Transcoding With Load Balancing On Cloud Computing Song Lin, Xinfeng Zhang, Qin Y, Siwei Ma Circuits and Systems, 2013 IEEE

Introduction #5

Paralleling in GOP-based

Thread1

Thread2

Thread3

Page 8: Parallelizing Video Transcoding With Load Balancing On Cloud Computing Song Lin, Xinfeng Zhang, Qin Y, Siwei Ma Circuits and Systems, 2013 IEEE

Related work #1

FCFS - First come first server [2]Easy to implementLoad balancing problem is still exist

Page 9: Parallelizing Video Transcoding With Load Balancing On Cloud Computing Song Lin, Xinfeng Zhang, Qin Y, Siwei Ma Circuits and Systems, 2013 IEEE

Related work #3

MCT – Minimal complete time [6]Map-Reduce-based

Page 10: Parallelizing Video Transcoding With Load Balancing On Cloud Computing Song Lin, Xinfeng Zhang, Qin Y, Siwei Ma Circuits and Systems, 2013 IEEE

Problem formulation and system architecture #1 Load balance problem on cloud computing

Executing timeDelay time

Data passing

C is complexity and P is computing capacity

Page 11: Parallelizing Video Transcoding With Load Balancing On Cloud Computing Song Lin, Xinfeng Zhang, Qin Y, Siwei Ma Circuits and Systems, 2013 IEEE

Problem formulation and system architecture #2 The overall completion time of set Sk is

.

Goal .

Page 12: Parallelizing Video Transcoding With Load Balancing On Cloud Computing Song Lin, Xinfeng Zhang, Qin Y, Siwei Ma Circuits and Systems, 2013 IEEE

Problem formulation and system architecture #3 Optimal solution

.

n means n task and m means m cores

Page 13: Parallelizing Video Transcoding With Load Balancing On Cloud Computing Song Lin, Xinfeng Zhang, Qin Y, Siwei Ma Circuits and Systems, 2013 IEEE

Problem formulation and system architecture #4 Flow chart of the proposed method

Page 14: Parallelizing Video Transcoding With Load Balancing On Cloud Computing Song Lin, Xinfeng Zhang, Qin Y, Siwei Ma Circuits and Systems, 2013 IEEE

Problem formulation and system architecture #5For video coding, if the input sequence has

instantaneous decoder refresh (IDR) frame, this video coding task can be divided into sub-tasks.[7]

Page 15: Parallelizing Video Transcoding With Load Balancing On Cloud Computing Song Lin, Xinfeng Zhang, Qin Y, Siwei Ma Circuits and Systems, 2013 IEEE

Problem formulation and system architecture #6For complexity estimation of video transcoding

tasks, the existing algorithms [8] [9] can be utilized.

Page 16: Parallelizing Video Transcoding With Load Balancing On Cloud Computing Song Lin, Xinfeng Zhang, Qin Y, Siwei Ma Circuits and Systems, 2013 IEEE

Proposed method #1

The framework includes three modulesTask pre-analysisAdaptive threshold segmentationMinimal finish time

Page 17: Parallelizing Video Transcoding With Load Balancing On Cloud Computing Song Lin, Xinfeng Zhang, Qin Y, Siwei Ma Circuits and Systems, 2013 IEEE

Proposed method #2

The threshold of segmentation

Page 18: Parallelizing Video Transcoding With Load Balancing On Cloud Computing Song Lin, Xinfeng Zhang, Qin Y, Siwei Ma Circuits and Systems, 2013 IEEE

Proposed method #3

Page 19: Parallelizing Video Transcoding With Load Balancing On Cloud Computing Song Lin, Xinfeng Zhang, Qin Y, Siwei Ma Circuits and Systems, 2013 IEEE

Proposed method #4

The optical finish time

The finish time

Page 20: Parallelizing Video Transcoding With Load Balancing On Cloud Computing Song Lin, Xinfeng Zhang, Qin Y, Siwei Ma Circuits and Systems, 2013 IEEE

Proposed method #5

Assign all the tasks sequentially in descending complexity order

For each unassigned task j, the cores are judged in their descending computing capacity order according to the following criterion: assuming the task j is assigned to core k, if Τκ ≤ Tthr, the assignment is verified. Otherwise, we will judge the next core.

Page 21: Parallelizing Video Transcoding With Load Balancing On Cloud Computing Song Lin, Xinfeng Zhang, Qin Y, Siwei Ma Circuits and Systems, 2013 IEEE

Proposed method #6

If all the cores are traversed and all the computing time are beyond Tthr, the task j will be assigned by MCT algorithm. and Tthr is updated to be the new finish time of the received core Tk

Page 22: Parallelizing Video Transcoding With Load Balancing On Cloud Computing Song Lin, Xinfeng Zhang, Qin Y, Siwei Ma Circuits and Systems, 2013 IEEE

Proposed method #7

Page 23: Parallelizing Video Transcoding With Load Balancing On Cloud Computing Song Lin, Xinfeng Zhang, Qin Y, Siwei Ma Circuits and Systems, 2013 IEEE

Experiment results #1

Page 24: Parallelizing Video Transcoding With Load Balancing On Cloud Computing Song Lin, Xinfeng Zhang, Qin Y, Siwei Ma Circuits and Systems, 2013 IEEE

Experiment results #2

Page 25: Parallelizing Video Transcoding With Load Balancing On Cloud Computing Song Lin, Xinfeng Zhang, Qin Y, Siwei Ma Circuits and Systems, 2013 IEEE

Experiment results #3

Page 26: Parallelizing Video Transcoding With Load Balancing On Cloud Computing Song Lin, Xinfeng Zhang, Qin Y, Siwei Ma Circuits and Systems, 2013 IEEE

Conclusion

Load balancing problem is a NP-hard problem.The proposed algorithm has strong robustness to

the task launching delay.