Abstract: Objects are cached in a system adapted to receive requests for said objects from one or more clients. The system has at least one of a plurality of nodes. Each node of the plurality of nodes includes at least one of a plurality of caches. One or more performance criteria of the system are selected. At least one of a plurality of metrics is assigned to a selected performance criterion. The desirability of storing an object of the objects in any one of the plurality of caches is calculated by assigning at least one of a plurality of desirability values to at least one of the objects based on the at least one of the plurality metrics. Any one of the objects stored in the at least one of the plurality of caches is replaced based on the at least one of the plurality of desirability values.

Abstract: A clustered computer system includes a shared data storage system, preferably a virtual shared disk (VSD) memory system, to which the computers in the cluster write data and from which the computers read data, using data access requests. The data access requests can be associated with deadlines, and individual storage devices in the shared storage system satisfy competing requests based on the deadlines of the requests. The deadlines can be updated and requests can be killed, to facilitate real time data access for, e.g., multimedia applications such as video on demand.

Abstract: In a client-server environment having a server group including one or more participating servers and one or more controller servers, where each participating server can communicate with at least one controller server, a method for coordinating actions among the servers in the server group comprising the steps of: (1) communicating between a client and at least one of the servers in the server group using a protocol that has limited or no defined procedures for passing state information between the client and the servers; (2) transmitting a request for an action from the client to one of the servers; and (3) communicating among a plurality of servers with at least one controller server coordinating actions of at least some of the servers in response to the request.

Abstract: A computerized method for load balancing in a geographically distributed or clustered system is disclosed. An arbiter assigns clients to nodes. The arbiter partitions clients into groups based on their request load. Each group is dynamically scheduled among nodes, thus avoiding high load groups from being allocated to the same node and overloading the system. If one of the nodes becomes overload, an alarm is generated, so that fewer or no new clients are allocated to the overloaded node.

Abstract: A method and system for recovering from a failure of a processing node in a partitioned shared nothing database processing system are provided. The processing system may include a pair of processing nodes having twin-tailed-connected thereto a storage device. A first processing node of the pair of processing nodes has a first database instance running thereon which accesses a first data partition on the storage device prior to the failure. Upon detection of the failure, access to the first data partition on the storage device is provided to a third, spare processing node through the second processing node of the pair of processing nodes. The third processing node runs a replacement database instance for the first database instance which was running on the first processing node prior to the failure thereof. The replacement database instance accesses the first data partition on the storage device through the second processing node, thereby recovering from the failure of the first processing node.

Abstract: A system and method for a general and extensible infrastructure providing monitoring and recovery of interdependent systems in a distributed/clustered system is disclosed. Subsystems, built without provision for high availability, are incorporated into the infrastructure without modification to core subsystem function. The infrastructure is comprised of one or more computing nodes connected by one or more interconnection networks, and running one or more distributed subsystems. The infrastructure monitors the computing nodes using one or more heartbeat and membership protocols, and monitors the said distributed subsystems by subsystem-specific monitors. Events detected by monitors are sent to event handlers. Event handlers process events by filtering them through activities such as event correlation, removal of duplicates, and rollup.

Abstract: A method and system are provided for partitioning a database and for balancing the processing load among processing nodes in a data processing system. A database partition is stored at each node and a logical owner node is assigned to each partition. The logical owner becomes responsible for all database requests related to the corresponding partition. Techniques for optimally assigning logical owners to all partitions in a system are provided and are responsive to parameters such as the load at each node, the combined load of the system, and the total number of assignments allowed. Techniques are also provided to accomplish the transfer of partition ownership between processing nodes when multiple transactions related to the partition are simultaneously occurring.