Horizontal architecture for high performance parallel communication systems

With the advent of high-speed networking technologies such as fiber optics, a traditional bottleneck in communication has disappeared. Now, the speed at which a processor can execute a communication protocol is the most significant limiting factor in protocol performance. The question is, how can a multimedia application take best advantage of these high-speed networks? We adopt an optimized implementation approach that allows efficient use of the bandwidth available on today´s high-speed networks. In this paper, we propose a new horizontal processing architecture that processes data as soon as it arrives from the network. This model increases the processing speed of the OSI model by using a multiple-instruction, single-data (MISD) scheme. However, the OSI protocol stack often imposes ordering constraints that prevent concurrent processing of the protocol layers. By using a fixed packet format, all layers from the network layer through to the presentation layer are processed in parallel. A prototype shows that the proposed model has a performance improvement of up to 60% more than the conventional approach