ABSTRACT
In this chapter we propose a dynamic hybrid scheduling [39], in which any new request for a pull item is kept in the pull queue. However, the clients’ impatience resulting from prolonged waiting for any item, or a new request for the same data item by another client often makes them transmit repeated requests. The server keeps these repeated requests in the repeat attempt (retrial) queue, thereby distinguishing such requests from the new requests arriving in the pull queue. At any instant of time the item to be serviced is selected by using stretch (i.e, max-request min-service-time first) optimal scheduling algorithm. The service of an item from the pull queue needs to consider the service of the instances of same items from the repeat attempt queue also. Using a multidimensional Markov model, the average performance of the overall heterogenous, hybrid scheduling system is derived.
