Method and apparatus to enable servers to manage long running asynchronous requests (03-Nov-2009)

Method and apparatus to enable servers to manage long running asynchronous requests

In a distributed computing environment it is often desirable to allow clients to submit asynchronous requests, making their application seem more responsive as they can carry out other tasks while waiting on requests to be completed. However, when the data in applications increases and the time taken to complete tasks increases this asynchronous behaviour can allow the customer to resubmit long running tasks numerous time dramatically increasing the workload on the server. Technologies like AJAX have tried to resolve this problem, however it expects the server to continually poll the HTTP request object to verify that the client is still connected, which is far from ideal from a performance perspective (does not scale terribly well).

    This invention solves this problem by categorizing and tracking requests submitted by a particular application. In one embodiment the server would receive a request; determine the priority of the request; interrupt any existing workload, and prune any queued requests.

    This invention works by adding information to the requests that allows the server to identify who is sending requests. When a new request is submitted the server could check if that user had any long running, interruptible, requests currently being processed. If a request is currently being executed and it is interruptible, then that request would...