Abstract: A distributed computing network has a packet routing system for routing packets among packet lists accessible to tasks. Packets are routed and attached to lists identified within a packet header or selected from router tables. Access to packet lists is managed according to data stored within packet list headers. Several types of packets are used for exchanging data, executing remote instructions, maintaining communication between network nodes, and supplying initialization instructions to network nodes. An inductive booting mechanism uses the packet routing system to initialize nodes and to re-boot failed nodes. Interface tasks manage transmission and reception of packets between nodes in order to provide asynchronous communication between tasks in a multitasking distributed computing environment.
Abstract: A hierarchical data structure is provided in the form of a tree for storing key-indexed entries. Each entry of the tree includes a balance bias indicator and pointer fields for maintaining (i) link pointers to successive entries in the hierarchy or (ii) thread pointers to preceding entries in the hierarchy. Entries are added to the tree, or removed from the tree, while the tree is maintained in a height-balanced condition. During insertion of a new entry, the tree is traversed along a search path to determine an insertion point for the entry and to determine a potential rebalancing point in the tree. The potential rebalancing point in the tree is identified on the basis of the balance bias indicators of the entries in the search path. The tree is rebalanced, if necessary, about the potential rebalancing point determined during the insertion traversal.