230 likes | 360 Views
OMFS An Object-Oriented Multimedia File System for Cluster Streaming Server. CHENG Bin, JIN Hai Cluster & Grid Computing Lab Huazhong University of Science & Technology HPCAsia, BeiJing , China , Dec. 200 5. Outline. Background Our Object-Oriented Method Improvement Schemes
E N D
OMFSAn Object-Oriented Multimedia File System for Cluster Streaming Server CHENG Bin, JIN Hai Cluster & Grid Computing Lab Huazhong University of Science & Technology HPCAsia, BeiJing, China, Dec. 2005
Outline • Background • Our Object-Oriented Method • Improvement Schemes • Performance Evaluation • Conclusion
Background From Centralized Streaming Server to Cluster Streaming Server.
Cluster File System Background • Existing cluster streaming servers • I/O Bottleneck • Poor Scalability • Block-based File Operations • Have to care about the media file format
Background • Existing multimedia file systems • Tiger Shark, PVFS • A general cluster file system HERMES, Symphony • focus on a QoS-aware disk scheduling algorithm, data placement, and cache policies. EXT3NS ( MM’05) • a local multimedia file system, not special for cluster streaming server
Background • The features of media files • Various of file formats, such mp4, mov, avi • Different files have different objection organizations • Be consist of logical object units • The features of cluster streaming server • Only perform reading operations • Seek objections according to timestamp. • Retrieve objects, such as meta object, data objects
Our goals • An object-oriented multimedia file system specially designed for cluster streaming server • To provide an object-oriented method to fetch the media data • To present a single system image • To mask differences of multimedia file format • To separate the processing of multimedia files from cluster streaming servers • To enhance the performance of cluster streaming server.
Object Oriented Method • Traditional processing method Cluster Streaming Server avi Parser_avi RTSP Server consume a lot of CPU and memory resource, not efficient mov Parser_mov RTP Server rm Parser_rm Parsing Block Object
Object Oriented Method Cluster Streaming Server Pre-processing Uniform Interface DB avi Parser_avi Meta Objects RTSP Server Apply the rule: Make the common case faster mov Parser_mov RTP Server rm RTP Ojbects Parser_rm Reading
Data Server DataSend Command Kernel Module 2 1 NIC DISK Bypassing Optimization Schemes • Bypassing kernel buffer Data Server 3 2 4 1 NIC DISK
Optimization Schemes • Creating time-based block indexing map Map Write Read Data Object Data Object Data Object Data Object Key Frame
Performance Evaluation • Simulation Setup • Our cluster streaming server has one Web Server, one Control Server, eight Data Servers, and one OMFS admin node. ---Two 1.4 GB AMD Opteron CPU ---2 GB Memory ---1000Mb/s Network Card ---avi, mp4, wmv Format • A multithreads simulator to generate requests, which accord with Poisson Distribution.
Performance Evaluation • Metrics • Max Concurrent Stream Number • Max Throughput • CPU Overhead • Average VCR Waiting Time
Simulation Results • Max Concurrent Stream Number about 12% improvement
Simulation Results • Max Throughput
Simulation Results • CPU Overhead
Simulation Results • Delay of VCR Operation
Conclusion • OMFS has the following features: • Simplify the design of cluster streaming server • Improve its performance • optimization schemes have great positive effect to the performance improvement of cluster streaming server.
Uniform Interface OMFS_Write(int object_type, void * object_content) OMFS_Read(int object_type, void * object_content)