Radisys Corporation

Oresis Communications

Andrew P Alleman

Michael W Hu

George Shenoda

Stateless load balancing of network packets within a system avoids detection by a network client or end user for deep packet inspection or other bump-in-the-wire applications. At least one header field of a received packet is used in generating a hash value. The hash value is used to identify a processing resource within the system for processing the received packet. Before being sent to the identified resource, the received packet is encapsulated with a new header that includes an indication of ingress port. The encapsulation does not modify the original packet. On a return path from the identified processing resource, the ingress port is determined from the encapsulated packet, the encapsulated packet is decapsulated to obtain a recovered packet that is identical to the received packet, and the recovered packet is forwarded to the network through an egress port as determined from the recovered ingress port.

Stateless load balancer in a multi-node system for transparent processing with packet preservation

Backplane redundancy is provided for a system including multiple nodes that communicate packets through first and second switches. Assuming that the first switch is initially assigned to an active state and the second switch to a standby state, the nodes communicate the data packets through physically enabled first backplane links to the first switch. The nodes physically enable second backplane links that are in a condition to communicate the data packets to the second switch. A messageless failover process is initiated by temporarily disabling, at the first switch, the first backplane links between the first switch and the nodes. In response to the nodes detecting the disabled first backplane links to the first switch, the nodes reconfigure themselves to communicate the data packets through the second backplane links to the second switch and to stop communicating the packets through the first backplane links to the first switch.

Switch redundancy in systems with dual-star backplanes

The distributed architecture switch of the invention provides a scalable, flexible switching device to route telephone calls and other data (e.g., frame relay, ISDN) over an asynchronous transfer mode (ATM) network. The switch includes multiple service modules that can be geographically disparate and function as a single switching device. The multiple service modules are coupled to a system controller that controls the service modules and passes messages between the service modules.

Growing community of inventors

Portland, OR, United States of America

Andrew P Alleman