Abstract: A device, method, and computer readable storage medium for routing in a computational grid arranged in a fat-tree, including performing preprocessing, performing base path routing to determine rules for routing in switches based on stored switch level, switch group, links between switches, links between switches and hosts, and fat-tree type, and performing run-time processing, including detecting a path fault, and determining at least one redundant path for the path fault. The base path routing determines a first routing rule that specifies a certain destination in every switch. The run-time processing further performs searching for a first match between a packet and a source-destination pair, and when the first match is not found, performs searching for a second match between the packet and the first routing rule that specifies the certain destination. By using routing rules that specific a destination, the routing in a computational grid can significantly scale up.

Abstract: A device and method for providing balanced routing paths in a computational grid including determining a type of topology of the computational grid having a plurality of levels, wherein each level includes a plurality of switches, determining whether the type of topology of the computational grid is a fat-tree, determining whether the fat-tree is odd, determining whether the fat-tree is a regular fat-tree, computing a first set of routing paths for the computational grid based on the determining of whether the fat-tree is odd and is a regular fat-tree, computing a second set of routing paths for the computational grid using a topology agnostic routing technique, and configuring forwarding tables in said switches with the first set of computed routing paths when the topology is determined to be a fat-tree and with the second set of computed routing paths when the topology is determined to not be a fat-tree.

Abstract: A device, method, and computer readable storage medium for routing in a computational grid arranged in a fat-tree, including performing preprocessing, performing base path routing to determine rules for routing in switches based on stored switch level, switch group, links between switches, links between switches and hosts, and fat-tree type, and performing run-time processing, including detecting a path fault, and determining at least one redundant path for the path fault. The base path routing determines a first routing rule that specifies a certain destination in every switch. The run-time processing further performs searching for a first match between a packet and a source-destination pair, and when the first match is not found, performs searching for a second match between the packet and the first routing rule that specifies the certain destination. By using routing rules that specific a destination, the routing in a computational grid can significantly scale up.

Abstract: A device and method for providing balanced routing paths in a computational grid including determining a type of topology of the computational grid having a plurality of levels, wherein each level includes a plurality of switches, determining whether the type of topology of the computational grid is a fat-tree, determining whether the fat-tree is odd, determining whether the fat-tree is a regular fat-tree, computing a first set of routing paths for the computational grid based on the determining of whether the fat-tree is odd and is a regular fat-tree, computing a second set of routing paths for the computational grid using a topology agnostic routing technique, and configuring forwarding tables in said switches with the first set of computed routing paths when the topology is determined to be a fat-tree and with the second set of computed routing paths when the topology is determined to not be a fat-tree.

Abstract: A system and method can provide fault tolerance in a middleware machine environment. A subnet manager can determine whether there is a path record change when a fault occurs in the middleware machine environment. Furthermore, the subnet manager can signal a dynamic reconfiguration event to at least one host in the middleware machine environment. The at least one host can send a message to the subnet manager to query for a latest path record. Then, the subnet manager can provide a latest path record to the at least one host.