Abstract: Improving scalability and throughput in publish/subscribe messaging systems by allowing the spoke systems to vary the member of the hub collective that it attaches to. The scalability is improved because the spoke systems do not have to have a long lasting attachment to a specific hub system. Instead, the spoke systems can be redistributed more evenly when new spoke systems are added or removed. The throughput is increased by distributing workload and fan-out responsibilities within hub collectives. Availability is improved because single points of failure are removed. Ease of administration is also improved over a hierarchy by allowing child nodes to automatically change parent nodes if a parent node is modified or removed.

Abstract: A sequence of processing transactions operating on a cached data set, which depend on each other because they operate on the same items of data from the data set. The transactions are saved until the master copy of the data becomes available. The transactions are played back against the master copy until one of the played back transactions fails because another transaction which operated on the same items of data but against a separate cached copy and originating from another application, was successfully played back on the master copy at an earlier time. At this point, all subsequent transactions which depended on the failing transaction are rolled back on the cached copy (with appropriate failure reports generated) and activity resumes from there. “Chained” (optimistic) transactions can therefore be applied to cached data and can be allowed to progress by recording their dependencies so they can be undone when required.

Abstract: Contention in a distributed processor computer system with a replicated message environment is reduced. The system comprises processor consumers intercommunicating via a server interface that receives a message from a producer. The server favors one consumer, marks the message accordingly, communicates the marked message to the system, and delays the processing in another consumer beyond a predetermined time, for example the time normally taken to process a message.

Abstract: Improving scalability and throughput in publish/subscribe messaging systems by allowing the spoke systems to vary the member of the hub collective that is attaches to. The scalability is improved because the spoke systems do not have to have a long lasting attachment to a specific hub system. Instead, the spoke systems can be redistributed more evenly when new spoke systems are added or removed. The throughput is increased by distributing workload and fan-out responsibilities within hub collectives. Availability is improved because single points of failure are removed. Ease of administration is also improved over a hierarchy by allowing child nodes to automatically change parent nodes if a parent node is modified or removed.

Abstract: A sequence of processing transactions operating on a cached data set, which depend on each other because they operate on the same items of data from the data set. The transactions are saved until the master copy of the data becomes available. The transactions are played back against the master copy until one of the played back transactions fails because another transaction which operated on the same items of data but against a separate cached copy and originating from another application, was successfully played back on the master copy at an earlier time. At this point, all subsequent transactions which depended on the failing transaction are rolled back on the cached copy (with appropriate failure reports generated) and activity resumes from there. “Chained” (optimistic) transactions can therefore be applied to cached data and can be allowed to progress by recording their dependencies so they can be undone when required.

Abstract: A sequence of processing transactions operating on a cached data set, which depend on each other because they operate on the same items of data from the data set. The transactions are saved until the master copy of the data becomes available. The transactions are played back against the master copy until one of the played back transactions fails because another transaction which operated on the same items of data but against a separate cached copy and originating from another application, was successfully played back on the master copy at an earlier time. At this point, all subsequent transactions which depended on the failing transaction are rolled back on the cached copy (with appropriate failure reports generated) and activity resumes from there. “Chained” (optimistic) transactions can therefore be applied to cached data and can be allowed to progress by recording their dependencies so they can be undone when required.

Abstract: A data processing method of testing a network of data processing units includes steps of sending a test message from an originating data processing unit to a destination data processing unit via an intermediate data processing unit; receiving at the originating data processing unit a channel definition from the intermediate data processing unit, the channel definition defining an action taken at the intermediate data processing unit to determine a next intermediate data processing unit to forward on the test message to; and displaying the channel definition at the originating data processing unit.

Abstract: A data processing method of testing a network of data processing units includes steps of sending a test message from an originating data processing unit to a destination data processing unit via an intermediate data processing unit; receiving at the originating data processing unit a channel definition from the intermediate data processing unit, the channel definition defining an action taken at the intermediate data processing unit to determine a next intermediate data processing unit to forward on the test message to; and displaying the channel definition at the originating data processing unit.

Abstract: A method of communicating packet based digital data over a communications network between a sender and a receiver, includes steps of sending a request, the request having a plurality of packets, to a receiver; receiving an acknowledgement from the receiver for each packet of the request; receiving a reply from the receiver, the reply having a plurality of packets, at least one of the packets including an indication that the packet need not be acknowledged upon receipt; and acknowledging only the packets of the reply that do not contain the indication.

Abstract: A data processing apparatus for carrying out workload management with respect to a group of servers in an asynchronous client/server computing system, has: a device for receiving a work request from a client application; a device for consulting availability data concerning the group of servers and for selecting a particular server in the group to be used to satisfy the received work request on the basis of the consulted availability data, where the consulted availability data is always used in its present state when the work request is received, irrespective of whether availability data has been received for each server in the group, in order to select a server from the group; and a device for placing the received work request in a transmission queue where it awaits transmission to the particular server in the group.