Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for distributed data management. One of the methods includes maintaining, by a first member in a distributed data management system having multiple computing members installed on multiple respective computers, a first garbage collection version vector that includes, for each member in the distributed data management system, a garbage collection version that represents a number of garbage collection processes performed by the member on a respective copy of a replicated data region maintained by the member in the data management system. If the first garbage collection version vector is different than a second garbage collection version vector received from a different provider member, a first replication process is performed that is different than a second replication process that is performed when the first garbage collection version vector matches the second garbage collection version vector.

Abstract: Methods, systems, and computer readable media for backing up and restore a deployment on a cloud computing platform are described. A deployment on a cloud computing platform includes a set of virtual machines and corresponding disks. A deployment may include multiple subunits designated as deployment jobs. A respective job specification file specifies respective resources used by each deployment job. The resources include various scripts specifying actions to take before, during, and after a backup. A backup orchestrator, upon receiving a backup request for the deployment, executes the scripts in a pre-specified sequence ordered by script type. When a particular type of scripts executes, the corresponding script for each deployment job, if specified in the specification file to correspond to that type, executes. When all scripts of a particular type finish execution, scripts having a next type execute. Accordingly, the operations of backing up multiple deployment jobs are synchronized and orchestrated.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for creating a probabilistic healthcare value profile for a patient.

Abstract: A computer readable storage medium comprises executable instructions to receive a query. A graph is built to represent jobs associated with the query. The jobs are assigned to parallel threads according to the graph.

Abstract: This specification describes methods, systems, and computer program products for maintaining data representing where each data block of multiple data blocks are stored among multiple computing nodes. Each computing node generates a respective locality summary based on locally stored data blocks, and submits the locality summary to a controlling computing node.

Abstract: Data accessible to a first process is replicated for use by a second process. Modifications to the data during the replication process may cause algorithms requiring the values of data both before and after replication to pause. Sending the values of the datum, before and after modification, to a process that will access the replicated data enables algorithms that use both values to initiate execution without waiting for the replication process to be completed.

Abstract: A deployment system provides the ability to deploy a multi-node distributed application, such as a cloud computing platform application that has a plurality of interconnected nodes performing specialized jobs. The deployment system may update a currently running cloud computing platform application according to a deployment manifest and a versioned release bundle that includes jobs and application packages. The deployment system determines changes to the currently running cloud computing platform application and distributes changes to each job to deployment agents executing on VMs. The deployment agents apply the updated jobs to their respective VMs (e.g., launching applications), thereby deploying an updated version of cloud computing platform application.

Abstract: In a mirrored database system, a careful write of intentions to perform file system actions is recorded in a persistent file system objects table that is flushed to disk prior to the actions being taken. The table durably and accurately records identities of file system objects that were in use by the database to facilitate creation and deletion of physical file directories and files on a database during crash recovery and during mirror resynchronize. In the event of a failure, crash recovery may quickly and easily identify file system objects which need to be cleaned up by reference to the persistent file system objects table. Similarly, resynchronization of the mirror database can be performed quickly by referring to the persistent file system table data to detect changes since the last database checkpoint.

Abstract: Methods, systems, and computer-readable media for monitoring states of application packages deployed on a cloud-based application deployment platform. A notification service retrieves a copy of a deployed application package and metadata associated with the application package from the cloud-based deployment platform, and identifies libraries of the application package. The notification service can then determine which, if any, libraries are or will become out-of-date, and obtain license information about the libraries. The notification service can provide notifications of any outdated components and on license compatibilities or incompatibilities. The notification service can automatically restage the application package, or update the license, upon finding outdated components or license incompatibilities.

Abstract: A distributed parallel processing database that processes data in a Java environment allocates memory both on a Java heap and off a Java heap. The distributed parallel processing database includes multiple servers. Each server executes a Java virtual machine (JVM) in which data allocated to the server is processed. When a JVM of a server starts, the JVM can specify an off-heap memory size, based on a JVM start parameter. The server can designate memory of the specified size that is off JVM memory heap as off-heap memory. The off-heap memory is different from heap memory in the Java environment, and is managed by a garbage collector that is outside of the Java environment. The server can process data designated as off-heap memory eligible in the off-heap memory. The off-heap memory can improve database operations that create a large number of similar-sized objects in memory by reducing Java memory management overhead.

Abstract: Methods, systems, and computer-readable media for on-demand resource provisioning for service instances. An on-demand service broker provisions IaaS resources at service instance creating time. The service broker provides a catalog listing one or more service plans, each service plan corresponds to a set of available resources. A user device selects a respective service plan that matches specific needs of an application that consumes the resources. The service broker generates a deployment manifest based on the selected service plan. The service broker then submits the deployment manifest to a deployment system. The deployment system provisions the resources to the service instance according to the manifest generated by the service broker.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for classifying events in a stream computing system using hierarchical analytic models. One of the methods includes receiving, by a stream computing system, data representing the values of one or more data attributes of an event in a stream of events. The values of each of the one or more data attributes are evaluated according to respective attribute-specific and class-specific criteria of a hierarchical analytic model in a predetermined order defined by the model. When a first value of a first data attribute satisfies one or more particular criteria for a first class, the first class of the plurality of classes is assigned to the event.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for performing semi-supervised learning on partially labeled nodes on a bipartite graph. One described method can determine a useful score of malware infection risk from partial known facts for entities modeled as nodes on a bipartite graph, where network traffic is measured between inside-the-enterprise entities and outside-the-enterprise entities. This and other methods can be implemented in a large-scale massively parallel processing database. Methods of scaling the partial label input and of presenting the results are also described.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for distributed data management. One of the methods includes maintaining, by a first member in a distributed data management system having multiple computing members installed on multiple respective computers, a first garbage collection version vector that includes, for each member in the distributed data management system, a garbage collection version that represents a number of garbage collection processes performed by the member on a respective copy of a replicated data region maintained by the member in the data management system. If the first garbage collection version vector is different than a second garbage collection version vector received from a different provider member, a first replication process is performed that is different than a second replication process that is performed when the first garbage collection version vector matches the second garbage collection version vector.

Abstract: A mobile computing device is operated to receive a trigger at a first instance. The trigger may be associated with a predefined condition or event or action. The mobile computing device may detect the predefined condition or event at a second instance. In response to detecting the predefined condition or event, a notification is activated on the mobile computing device that is based on the trigger.

Abstract: SQL queries are optimized to operate directly on compressed data (and obtain the correct result) rather than requiring that the data be first decompressed prior to processing a query. Certain characteristic pattern trees are mapped against a logical input query plan that includes certain logical operators such as a DECOMPRESS that precedes a JOIN or a GROUPBY in association with a COUNT to identify instances in the plan that match a characteristic pattern. Upon locating a match, the input query plan is transformed into a logically equivalent plan that operates correctly on compressed data, by analyzing the interplay of the semantics of logical query operations with the compressed data and substituting less costly structures and operations. DECOMPRESS operations are moved to operate subsequent to a JOIN or eliminated altogether, and COUNT operations are replaced by a different operation, such as SUM, that is logically equivalent for compressed data.

Abstract: Query optimizer accuracy is tested by measuring the actual execution times on a database of each of a plurality of alternative query plans for each of a plurality of queries, first ranking the plans by the actual execution times, then ranking the plans by estimated execution times, and correlating the two rankings by computing a rank correlation metric. The rank correlation metric is normalized so that different optimizers can be objectively compared.

Abstract: In a mirrored database system, a careful write of intentions to perform file system actions is recorded in a persistent file system objects table that is flushed to disk prior to the actions being taken. The table durably and accurately records identities of file system objects that were in use by the database to facilitate creation and deletion of physical file directories and files on a database during crash recovery and during mirror resynchronize. In the event of a failure, crash recovery may quickly and easily identify file system objects which need to be cleaned up by reference to the persistent file system objects table. Similarly, resynchronization of the mirror database can be performed quickly by referring to the persistent file system table data to detect changes since the last database checkpoint.

Abstract: Systems, methods, and computer program products for trace management in a distributed computing environment are described. A trace manager receives a request to analyze a trace of a series of calls between applications serving requests in a microservice architecture. The trace manager determines metrics including, for example, latency information and start time for the series of calls. The trace manager presents the metrics in association with log entries correlated to the series of calls. A call stack, or one or more latency indicators, provides a user interface of selecting one or more calls. A user selection of a particular call in the call stack or a latency indicator causes one or more corresponding log entries to be emphasized or filtered. The call stack and the one or more latency indicators can have various display configurations.

Abstract: A random number generation process generated uncorrelated random numbers from identical random number sequences on parallel processing database segments of an MPP database without communications between the segments by establishing a different starting position in the sequence on each segment using an identifier that is unique to each segment, query slice information and the number of segments. A master node dispatches a seed value to initialize the random number sequence generation on all segments, and dispatches the query slice information and information as to the number of segments, during a normal query plan dispatch process.