Abstract: Methods and apparatus for hosting a network service on a cluster of servers, each including a primary and a secondary Internet Protocol (IP) address. A common cluster address is assigned as the secondary address to each of the servers in the cluster. The cluster address may be assigned in UNIX-based servers using an ifconfig alias option, and may be a ghost IP address that is not used as a primary address by any server in the cluster. Client requests directed to the cluster address are dispatched such that only one of the servers of the cluster responds to a given client request. The dispatching may use a routing-based technique, in which all client requests directed to the cluster address are routed to a dispatcher connected to the local network of the server cluster. The dispatcher then applies a hash function to the client IP address in order to select one of the servers to process the request.

Abstract: An optimistic distributed simulation method applicable to event-driven simulation that requires only a single rollback announcement per straggler message, with no need for other simulation processes to announce their roll backs. The method is accomplished by maintaining transitive dependency information between the simulation processes in the form of a simulation vector. The present invention eliminates the need for output queues and also the possibility of multiple roll backs per simulation process and avalanche of anti-messages. Alternatives to reduce blocking and minimize the size of the simulation vectors are also disclosed thereby reducing the overhead requirements for systems with large numbers of logical processes.

Abstract: An optimistic distributed simulation method applicable to event-driven simulation that requires only a single rollback announcement per straggler message, with no need for other simulation processes to announce their roll backs. The method is accomplished by maintaining transitive dependency information between the simulation processes in the form of a simulation vector. The present invention eliminates the need for output queues and also the possibility of multiple roll backs per simulation process and avalanche of anti-messages. Alternatives to reduce blocking and minimize the size of the simulation vectors are also disclosed thereby reducing the overhead requirements for systems with large numbers of logical processes.

Abstract: A fault tolerant message passing system includes a plurality of interconnected processors with storage and a watchdog process wherein the processors may undergo failure. A method restores a consistent system state using optimistic logging protocol with asynchronous recovery. Each process comprises a sequence of state intervals and includes checkpoints for saving in storage the state of the process sufficient to re-start execution of the process. Non-deterministic event messages are logged in storage by each process for replay after process re-start to reconstruct pre-failure state intervals. Transitive dependency tracking of messages and process states is performed to record the highest-index state interval of each process upon which a local process depends. A variable size dependency vector is attached to each outgoing message sent between processes. An integer K is assigned to each outgoing message as the upper bound on the vector size.