Method and apparatus of retrieving continuous and non-continuous media data from a file system

摘要

Retrieval of both continuous and non-continuous media data is performed concurrently for multiple requests, where servicing of continuous media data requests at varying rate requirements is guaranteed within a common retrieval period. The common period is selected with respect to the available buffer space and the total disk retrieval times required for servicing multiple requests. Servicing of requests is re-commenced immediately after all admitted requests have been serviced, regardless if the common period has elapsed. High throughput is obtained and transfer rates for a large number of real-time requests are guaranteed by reducing seek latency and eliminating rotational latency so that the buffering requirements for requests are reduced. Disk scheduling techniques are applied for disks having transfer rates which vary from one track to another. Non-continuous media requests are serviced concurrently with continuous media requests by reserving a certain portion of the disk bandwidth for non-continuous media requests in order to provide for low response times. Changing workloads are accommodated by dynamically varying the buffer space allocated to continuous media and non-continuous media requests and by using the time allocated but not utilized for continuous media requests for servicing non-continuous media requests. The disk scheduling techniques are also applicable for disks having varying track sizes and where clips are stored on tracks non-contiguously.