摘要:
A method and system is provided for tolerating component failure in a continuous media server system. The present invention guarantees data streams at a constant rate to subscribers for the data streams even when at least one component fails. The present invention is able to guarantee data streams at a constant rate by utilizing declustered mirroring and by reserving bandwidth for both normal mode processing and failure mode processing. The declustered mirroring of the present invention is performed by dividing the data to be stored in the continuous media server system into blocks. The blocks are then striped across the storage devices of the continuous media server system and each block is divided into a number of sub-blocks. The sub-blocks are in turn stored on separate storage devices. The present invention reserves bandwidth for both normal mode processing and failure mode processing. Since the present invention utilizes declustered mirroring, the bandwidth reserved for failure mode processing is reduced. Therefore, when a failure occurs, the bandwidth reserved for failure mode processing is utilized and the data streams to the subscribers are uninterrupted.
摘要:
Interface identifier parameterization techniques are described. In one or more implementations, a textual description is obtained of a parametric type represented in a binary standard of a programming model and an interface identifier is generated based at least in part on the obtained textual description of the parametric type.