Abstract: A computer management system may track services on multiple computers executing the same workload. Various performance counters may be retrieved from the computers and analyzed to identify those with low deviations and low mean deviations. Statistical tests may then be used to compare the values of a particular computer to the deviation of the group of computers to identify outliers, which may indicate degraded performance and imminent failure of the computer. The management system may analyze any collectable performance related parameter to identify outlying systems.

Abstract: A method for managing distributed computing. The method comprises estimating, for each of a plurality of local computing nodes, a distribution of multidimensional values in a space. Each multidimensional value is calculated according to a plurality of locally monitored parameters. The method further includes calculating safe zones in the space where each safe zone is defined according to a respective estimated distribution under a global geometric constraint in the space and setting local geometric constraints for the local computing nodes according to the respective safe zones. Each local geometric constraint is defined such that a detection of at least one monitored multidimensional value violating it by a respective local computing node induces a communication event between the respective local computing node and one or more central computing nodes.

Abstract: A computer management system may track services on multiple computers executing the same workload. Various performance counters may be retrieved from the computers and analyzed to identify those with low deviations and low mean deviations. Statistical tests may then be used to compare the values of a particular computer to the deviation of the group of computers to identify outliers, which may indicate degraded performance and imminent failure of the computer. The management system may analyze any collectable performance related parameter to identify outlying systems.

Abstract: A computer management system may track services on multiple computers executing the same workload. Various performance counters may be retrieved from the computers and analyzed to identify those with low deviations and low mean deviations. Statistical tests may then be used to compare the values of a particular computer to the deviation of the group of computers to identify outliers, which may indicate degraded performance and imminent failure of the computer. The management system may analyze any collectable performance related parameter to identify outlying systems.

Abstract: A method for distributed computing includes processing multiple sets of data at respective computing nodes (24), and calculating respective local values of one or more statistical parameters characterizing the sets of the data. A global condition is defined, such that the condition is violated when a function defined over a weighted average of the respective local values crosses a predetermined threshold. The global condition is separated into a plurality of local constraints, which include a respective local constraint to be evaluated by each of the nodes based on the respective local values, such that violation of the respective local constraint in at least one of the nodes indicates a violation of the global condition. The local constraint is evaluated independently at each of the nodes. When at least one of the nodes detects that the respective local constraint is violated, an indication that the global condition has been violated is produced.

Abstract: A computer management system may track services on multiple computers executing the same workload. Various performance counters may be retrieved from the computers and analyzed to identify those with low deviations and low mean deviations. Statistical tests may then be used to compare the values of a particular computer to the deviation of the group of computers to identify outliers, which may indicate degraded performance and imminent failure of the computer. The management system may analyze any collectable performance related parameter to identify outlying systems.

Abstract: A method for managing distributed computing. The method comprises estimating, for each of a plurality of local computing nodes, a distribution of multidimensional values in a space. Each multidimensional value is calculated according to a plurality of locally monitored parameters. The method further includes calculating safe zones in the space where each safe zone is defined according to a respective estimated distribution under a global geometric constraint in the space and setting local geometric constraints for the local computing nodes according to the respective safe zones. Each local geometric constraint is defined such that a detection of at least one monitored multidimensional value violating it by a respective local computing node induces a communication event between the respective local computing node and one or more central computing nodes.

Abstract: A method for discovering association rules in a distributed database that includes a plurality of partitions associated with respective computing nodes includes generating a candidate association rule defining an association relationship between itemsets in the distributed database. At each node among at least a subset of the nodes, an asynchronous fully-distributed majority voting process is applied to assess a correctness of the candidate association rule.

Abstract: A method for distributed computing includes processing multiple sets of data at respective computing nodes (24), and calculating respective local values of one or more statistical parameters characterizing the sets of the data. A global condition is defined, such that the condition is violated when a function defined over a weighted average of the respective local values crosses a predetermined threshold. The global condition is separated into a plurality of local constraints, which include a respective local constraint to be evaluated by each of the nodes based on the respective local values, such that violation of the respective local constraint in at least one of the nodes indicates a violation of the global condition. The local constraint is evaluated independently at each of the nodes. When at least one of the nodes detects that the respective local constraint is violated, an indication that the global condition has been violated is produced.

Abstract: A method for discovering association rules in a distributed database that includes a plurality of partitions associated with respective computing nodes includes generating a candidate association rule defining an association relationship between itemsets in the distributed database. At each node among at least a subset of the nodes, an asynchronous fully-distributed majority voting process is applied to assess a correctness of the candidate association rule.

Abstract: A method for mining association rules in a database that is divided into multiple partitions associated with respective computing nodes. The method includes transmitting messages among the nodes with respect to local support of an itemset in the respective partitions of the database. Responsive to the messages transmitted by a subset of the nodes, the itemset is determined to be globally frequent in the database before the nodes outside the subset have transmitted the messages with respect to the local support of the itemset in their respective partitions. An association rule is computed with respect to the itemset, responsive to having determined the itemset to be globally frequent.

Abstract: This disclosure presents a technique of field-level caching in distributed object-oriented systems, in which a speculative approach is taken to identify opportunities for caching. The speculative approach is particularly suitable for exploitation of opportunities for caching. Invalidation protocols, which are fully compliant with the Java memory model, are provided to recover from incorrect speculation, while incurring only a low overhead. The technique has been implemented on a cluster of machines, and has been found to be readily scalable with multi-threaded applications. Field caching, optionally combined with other optimizations, produces a practically important performance step up in distributed environments, such as the cluster virtual machine for Java, which transparently distributes an application's threads and objects among the nodes of a cluster.

Abstract: A method for controlling access to a physical memory in a distributed shared memory system (DSM) includes selecting a page of the physical memory in which to store a plurality of data items, including at least first and second data items. The selected page of the physical memory is divided into a plurality of minipages, including at least first and second minipages containing the first and second data items, respectively. First and second virtual pages are mapped to the selected page of the physical memory, such that the first and second virtual pages are associated respectively with the first and second minipages, with first and second access permissions applied to the first and second virtual pages, respectively. A process is permitted to access the first data item subject to the first access permission and the second data item subject to the second access permission, via the respective first and second virtual pages.

Abstract: A method for mining association rules in a database that is divided into multiple partitions associated with respective computing nodes. The method includes transmitting messages among the nodes with respect to local support of an itemset in the respective partitions of the database. Responsive to the messages transmitted by a subset of the nodes, the itemset is determined to be globally frequent in the database before the nodes outside the subset have transmitted the messages with respect to the local support of the itemset in their respective partitions. An association rule is computed with respect to the itemset, responsive to having determined the itemset to be globally frequent.

Abstract: This disclosure presents a technique of field-level caching in distributed object-oriented systems, in which a speculative approach is taken to identify opportunities for caching. The speculative approach is particularly suitable for exploitation of opportunities for caching. Invalidation protocols, which are fully compliant with the Java memory model, are provided to recover from incorrect speculation, while incurring only a low overhead. The technique has been implemented on a cluster of machines, and has been found to be readily scalable with multi-threaded applications. Field caching, optionally combined with other optimizations, produces a practically important performance step up in distributed environments, such as the cluster virtual machine for Java, which transparently distributes an application's threads and objects among the nodes of a cluster.

Abstract: A method for controlling access to a physical memory in a distributed shared memory system (DSM) includes selecting a page of the physical memory in which to store a plurality of data items, including at least first and second data items. The selected page of the physical memory is divided into a plurality of minipages, including at least first and second minipages containing the first and second data items, respectively. First and second virtual pages are mapped to the selected page of the physical memory, such that the first and second virtual pages are associated respectively with the first and second minipages, with first and second access permissions applied to the first and second virtual pages, respectively. A process is permitted to access the first data item subject to the first access permission and the second data item subject to the second access permission, via the respective first and second virtual pages.

Abstract: A Distributed Shared Memory (DSM) system that provides a single address space shared by a number of processes The DSM is capable of accomplishing lowered granularity of distributed shared memory. The DSM system includes the address space constituting a virtual address space that includes virtual pages. Two or more virtual pages in the virtual address space are mapped to the same memory page that includes one page. The page includes two or more minipages. There is further provided an assignor assigning independently, through the virtual pages, access permission to minipages that reside in the same page of the memory object. There is further provided processor processing independently in different processes minipages that reside in the same page.