International Business Machines Corporation

Alexei Pytel

Steven Simonson

Thomas Wasik

Kevin James Kathmann

Bruce Talbott

Lorne Parks

Gary Stuart Puchkoff

A technique relates to resource allocation of a countable managed resource. A computer sets a reserved amount of the managed resource to be available for a privileged status, such that there is an unreserved amount of the managed resource to be available for a non-privileged status. In response to a non-privileged job needing more of the unreserved amount of the managed resource than what is available, it is determined whether the non-privileged job is promotable to the privileged status. In response to the non-privileged job not being promotable to the privileged status, the non-privileged job is prevented from accessing the reserved amount of the managed resource. In response to the non-privileged job being promotable to the privileged status, the non-privileged job is permitted to utilize the reserved amount of the managed resource.

Prevention and resolution of a critical shortage of a shared resource in a multi-image operating system environment

A method, a computer program product, and a computer system for generating a set of scripts that perform processing equivalent to an original script and achieve maximum parallel executions. A computer receives an original script including annotations defined by a user, builds a graph comprising nodes and representing execution steps in the original script, associates one or more attributes to the nodes, and determines dependencies between the nodes, based on accessing data sets or files by the execution steps. The computer modifies the dependencies between the nodes, using instructions contained in the annotations. The computer generates a set of scripts that perform processing equivalent to the original script and achieve maximum parallel executions, based on the nodes of the graph and the dependencies. The computer generates a script that defines dependencies between the scripts.

Batch script decomposer

A hybrid messaging model including a method that sends a first request message from a control process executing on a computer to a plurality of subordinate processes. The first request message directs the subordinate processes to enter a first state. An expected state is set equal to the first state in response to sending the first request message. A status message, including the expected state, is periodically broadcast from the control process to the plurality of subordinate processes. At least one confirmation message is received from each of the subordinate processes confirming that the subordinate process is in the expected state. Each of the confirmation messages is responsive to either the first request message or to the status message. A second request message is sent from the control process to the plurality of subordinate processes in response to receiving at least one confirmation message from each of the subordinate processes.

Hybrid request/response and polling messaging model

A node container is constructed with multiple binary tree views. A processing device creates a primary view binary tree within the node container based on a primary key. A set of records that match fields of the primary key is chained to the primary view. The processing device also creates one or more alternate views within the node container to allow access to the same set of records via one or more alternate keys, which are subsets of the fields of the primary key. The chain of records is extracted from the node container when matched to a subsequent query search.

Dynamic output selection using highly optimized data structures

A method is provided for constructing a cache for storing results of previously evaluated queries in a binary tree based on a cache key. The cache is searched, by a processing device, for a node representing a set of previously evaluated queries that match a given record using an instance of the cache key. In response to finding a representative node that matches the instance of the cache key, the method performs processing associated with the set of previously evaluated queries. In response to finding no representative nodes, the method creates a new node using the instance of the cache key. The method of the invention then updates the cache by classifying the compatibility of newly registered queries with the cache key and evaluating whether compatible newly registered queries match the record.

Efficient selection of queries matching a record using a cache

Transparent file system migration including a method that identifies a source physical storage location in a file system that supports simultaneous access by multiple independent processes. A target physical storage location in the file system is identified and its contents are migrated to the target physical storage location. The migrating includes disallowing new allocations to the source physical storage location and copying contents from the source physical storage location to the target physical storage location. A write request to write data to the source physical storage location is received prior to or during the migrating. The write request is serviced contemporaneously with the migrating. Read and write requests directed to the source physical storage location are redirected to the target physical storage location. The redirecting is responsive to the migrating completing. The requests are from the multiple independent processes and include both logical and physical input/outputs.

Transparent file system migration to a new physical location

In a computer system that includes multiple processors, each processor in a computer system is assigned a processor class. Processor capacity and usage are monitored according to the class assigned to the processor. Capacity and usage are reported on a class-by-class basis so that the capacity and performance of different classes of processors are not erroneously compared or summed. The capacity and usage are monitored and reported in an abstract unit of measurement referred to as a 'CPU time unit'. Processors of the same type that run at different clock speeds or that have different internal circuitry enabled are preferably assigned the same class, with one or more conversion factors being used to appropriately scale the performance of the processors to the common CPU time unit for this class.

Growing community of inventors

Rochester, MN, United States of America

Alexei Pytel