Maximum receive capacity specifying query processing client/server system replying up to the capacity and sending the remainder upon subsequent request

Abstract

An Interactive Television (ITV) Client/Server system comprises one or more Clients (e.g., set top boxes or "STBs") connected to a Server module, via a Broadband Communication Network, is described. The system provides a generic mechanism for deciding how much or how little data is to be sent in response to a request from a Client. Specifically, each Client is allowed to describe ahead of time how much memory (approximately or exactly) it is allowing for the query result. This is communicated by the Client as part of the query itself. Regardless of what other activity the Client undertakes, the server maintains the context of the original query, until the Client terminates the connection. In a similar manner, when other simultaneous queries are open, the context is maintained for each query until the connection is dropped (or the query terminates). The Client can ask for more data (i.e., rows) on any open query. To optimize operation of the environment, requests from the Clients (i.e., queries) are de-multiplexed down to a set of worker threads, available at the server, which carry out the actual work of query processing. Because of this approach, a request for more data from a Client (i.e., set top box) can "land" on any worker thread. This gives queries the ability to "jump" from one thread of execution to another. In this manner, a finite number of threads may be employed to service a large query pool, with maximum throughput.

Dependencies are inferred from the claim text.