1 / 14

Continuous Media 1

Continuous Media 1. Differs significantly from textual and numeric data because of two fundamental characteristics: Real-time storage and retrieval High data transfer rate and large storage space. Continuous Media 2.

aquarius
Download Presentation

Continuous Media 1

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Continuous Media 1 • Differs significantly from textual and numeric data because of two fundamental characteristics: • Real-time storage and retrieval • High data transfer rate and large storage space

  2. Continuous Media 2 • Recording -- server continuously stores data produced by an input device to prevent buffer overruns at the device • Playback -- server retrieves data from the disk at a rate that prevents an output device from starving

  3. Single Stream Playback • A sequence of periodic tasks with deadlines • Tasks correspond to retrievals of media blocks from disk • Deadlines correspond to the scheduled playback times • Though just-in-time playback is conceivable, retrieval will be bursty • Media blocks need to be buffered when retrieval gets ahead of playback

  4. Challenge • To supply stream buffers with enough data to ensure that the playback processes do not starve • buffer entire stream before initiating playback • requires a large buffer and introduces lengthy latency • serve the stream periodically minimizing buffer space requirement and the initial latency

  5. Multistream Retrieval • Disk data transfer rates are significantly higher than a single stream’s playback rate • Retrieval requests from several streams simultaneously • Many of these streams may be accessing the same file, but different parts of the file at the same time • Continuous playback requirements of ALL streams must be met • Scheduling disk requests such that no individual stream starves and limiting the number of streams scheduled at a given time

  6. Disk Scheduling • Traditional algorithms -- First come, first served; Shortest seek time first; SCAN • to reduce seek time and rotational latencies, to achieve high throughput, to provide fair access • Real-time scheduling of tasks with deadlines -- EDF • schedules the media block with the earliest deadline • excessive seek time and rotational latency and poor server resource utilization

  7. Disk Scheduling for CM Data • SCAN-EDF -- serves the requests with earliest deadline first, but when several requests have the same deadline, their respective blocks are accessed with the SCAN algorithm • To improve the effectiveness of SCAN-EDF, increase the number of requests with the same deadline, for example one can enforce that all requests have deadlines that are multiples of a period

  8. Round Scheduling for CM Data • Process requests in rounds to exploit the nature of CM playback • Each round still requires a disk scheduling algorithm • Round-robin -- request order within a round is fixed and maximum latency between successive retrieval times is bounded by a round’s duration • SCAN -- request order depends on data placement and maximum latency between successive retrieval times is bounded by twice the round’s duration

  9. Round Length and Latency Trade-offs • Round-robin -- initial latency shorter but round lengths are longer • SCAN -- initial latency longer but round lengths are shorter • Grouped Sweeping Scheme (GSS) • partitions each round into groups • groups are served in fixed order • SCAN is used within each group

  10. Production versus Consumption • During a round, the amount of data retrieved for a stream is at least equal to the amount consumed by the stream’s playback • Data production never lags behind consumption (buffer conserving property) • Buffer conservation as a sufficient condition for preventing starvation • Maximum duration of a round, maximum consumption rate, same number of media blocks or proportional number of media blocks

  11. Admission Control • In admitting new streams, a multimedia server employs admission control tests to determine whether a new stream can be served without affecting the playback requirements of the streams that are already in service • Deterministic -- worst case scenarios and all deadlines are guaranteed to be met • Statistical -- Deadlines are guaranteed to be met with a certain probability • Background (or Best-effort) -- No guarantees are given and are served only if there is excess available capacity

  12. Data Placement • Retrieval Blocks • Contiguous Placement • a single seek is enough (but fragmentation is problem) • Scattered Placement • many intrafile seeks (can be alleviated using large block sizes and to read one block in each round) • Constrained Placement • intrafile seeks may be unavoidable if multiple blocks are needed in each round • reduce the seeks to a reasonable bound by bounding the separation of successive blocks

  13. Data Striping and Data Interleaving • RAID Scheme -- data is striped across each disk • achieves intrafile and interfile parallelism • Disk arrays are a good solution to the high bandwidth requirements of multimedia services • Data interleaving allows blocks of a media file to be stored on different disks as successive file blocks

  14. Multimedia File Systems • File system oriented • multimedia object as a large file and typical file system operations such as open, close, read are supported • Stream oriented • multimedia data continuously “streamed” to the client with operations such as play, pause, resume supported • File retrieval structures -- linked lists, FAT, indexes

More Related