Field Oriented Pipeline Architecture For A Programmable Data Streaming Processor

A distributed data processing system for executing database operations comprising:

• (a) a first group including at least one host processor configured to accept queries for data stored on mass storage devices;
• (b) a second group including two or more Job Processing Units (JPUs), each JPU configured to receive requests from the at least one host processor;
• (c) a network infrastructure providing network connection of the first and second groups as network nodes, to support communication of data among network nodes, the network infrastructure using a communication protocol;
• (d) a communication process responsible for communicating data blocks in one or more messages communicated among network nodes;
• (e) an application process for executing database operations, with portions of the application process executing on the JPUs, and portions of the application process executing on at least one host processor;
• and
• (f) the communication process implementing a load balancing function in front of the at least one host processors, the load balancing function directing individual transactions to selected ones of the at least one host processors so as to evenly distribute workload.