Abstract: A data-warehousing system allows various areas of an enterprise to view data at varying levels of data freshness. The system acquires data that represents an event in the life of a business enterprise, such as a transaction between the enterprise and one of its customers, and loads this data into a database table. The system then makes the data available for retrieval from the table and stores information indicating when the data was made available for retrieval. In some embodiments, the system also acquires data that is related to and more current than the data representing the event and stores the more current data in the database. The system then stores information indicating when the more current data was stored in the database. Such a data warehouse allows decision-makers in the business to see some information (e.g., customer transaction or account data) up-to-the-moment and other information as it stood at some specific point-in-time, such as at the end of the previous month.

Abstract: A method for spooling data for use in joining a small table with a large table in a relational database system. The method analyzes a join condition for combining records from the small and large tables, selects qualified rows from the large table, and writes the qualified rows to a spool file. The spool file includes a first partition containing hash values of all bind terms for the join condition; a second partition including a join column with a best selective bind term; and at least one additional partition including additional join columns used in bind terms. The partitions are grouped together within a container row in the spool file, and multiple container rows are written together within a super-container row in the spool file.

Abstract: A method for performing in-memory hash join processing. This method utilizes bulk processing within the hash join steps, such as performing bulk reads of hash values from tables to be joined, and performing bulk probes of hash values in tables to be joined, thereby providing more efficient utilization of memory bandwidth and CPU throughput, reducing memory accesses in the execution path, and reducing CPU cycles per instruction. Data movement is reduced by reducing load-stores to memory and by performing more operations in CPU cache.

Abstract: A database system comprises a plurality of processing modules arranged to process data objects from a plurality of data object servers based upon a database query from a client computer system. A control task module is arranged to iteratively dynamically assign data objects from the plurality of data object servers to each of the plurality of processing modules based upon processing activity associated with the processing module during the database query from the client computer system. Alternatively, or in addition to, the control task module is arranged to iteratively dynamically assign data objects from the plurality of data object servers to each of the plurality of processing modules based upon a characteristic of the data objects to be dynamically assigned to the processing module during the database query from the client computer system.

Abstract: Techniques for transitioning between code-based and data-based execution forms (or models) are disclosed. The techniques can be used to improve the performance of computing systems by allowing the execution to transition from one of the execution models to another one of the execution models that may be more suitable for carrying out the execution or effective processing of information in a computing system or environment. The techniques also allow switching back to the previous execution model when that previous model is more suitable than the execution model currently being used. In other words, the techniques allow transitioning (or switching) back and forth between a data-based and code-based execution (or information processing) models.

Abstract: A database system receives a request from a user. The request invokes a data set function (DSF) and uses a property to be provided by the DSF. The database system determines that a function descriptor is available for the DSF. The function descriptor is expressed as markup language instructions. The function descriptor defines the property of the DSF. The database system uses the function descriptor to define a property for the DSF.

Abstract: Execution of a query invoking an analytical function (AF) is optimized. The query includes a join operation between an AF table and an AuxiliaryTable and includes determining that the AF includes a plurality of AF properties. Query-level properties about the query are inferred. It is determined to change an order of the join operation from the plurality of AF properties and query-level properties.

Abstract: A method for detecting and handling skew and spillover in in-memory hash join operations. To improve the detection of skew and spillover in parallel processing systems, a Poisson distribution of unique hash values to Units of Parallelism (UoPs) is employed to determine the number of rows per UoP and in turn, the potential of spillover at a UoP. Hash join plan options can be selected or adjusted to reduce the likelihood of spillover.

Abstract: Multiple cost models (e.g., a sub-operations costing model and logical-operations costing model) can be used to make cost estimations of execution of database queries in one and each one of the multiple heterogeneous database systems. As a result, a “hybrid” cost estimating mode can be used whereby two or more cost models can be used in a single database system in to order maximize the advantages and minimize the disadvantages of each of the cost models, thereby striving to achieve an optimal balance. In addition, cost estimation can be switched between a hybrid cost estimating mode and a single cost estimating mode. The switch can, for example, be made as a part of tuning phase, as more information about actual costs of execution of database queries becomes more available, or as a result of changes to the database system and/or it operations, and so on. As a result, a flexible cost estimating mechanism can also be provided.

Abstract: Techniques for improving the execution of database queries in a multi-processor system or distributed processing system environment are provided. In a database system including multiple parsing engines (PEs) for parsing database queries, or requests, received by the system and generating execution plans for the requests, execution plans generated for requests can be saved in a global request cache accessible to each of the parsing engines. Requests which have been parsed and cached by a PE can be retrieved for use by other PEs, thereby avoiding unnecessarily parsing the same database request in multiple PEs. The global request cache may be a distributed cache consisting of request caches local to each parsing engine, with execution plans allocated to the local request caches using hashing techniques applied to the database requests associated with the execution plans.

Abstract: An apparatus, method and computer program product for neural network training over very large distributed datasets, wherein a relational database management system (RDBMS) is executed in a computer system comprised of a plurality of compute units, and the RDBMS manages a relational database comprised of one or more tables storing data. One or more local neural network models are trained in the compute units using the data stored locally on the compute units. At least one global neural network model is generated in the compute units by aggregating the local neural network models after the local neural network models are trained.

Abstract: Various embodiments herein each include at least one of systems, methods, and software for producing operational analytics that summarize fine-grain time scale behavior over long time durations. Some such embodiments are targeted toward understanding operationally meaningful behavior of complex dynamic systems that are often only apparent at fine-grain time scales. Such behavior occurs rarely and/or only for short durations so the analytics of some embodiments cover long time durations.

Abstract: Selecting a join plan for a query containing a join and a union block includes determining whether to propose a join plan with the join pushed across the union block. A selection is made between a join plan in which the join is not pushed across the union block and any proposed join plan in which the join is pushed across the union block.

Abstract: Data portions of a database can be grouped and ranked in order of priory for reassignment from one or more maps to another one or more maps. It should be noted that a first map can assign the data portions to a first configuration of processors for processing the data portions, and a second map assigns the data portions to a second configuration of processors, different than the first configuration, for processing the data portions in a database system and/or environment. The data portions are reassigned in groups during an available time (window) for reassignment by taking the first one of the groups can be reassigned (“moved”) in the available, then the second one in the available reaming time, and so on, until no group of data portions can be moved in the remaining time or all of them have been moved.

Abstract: A data store system includes a processor that may generate a hash value based on a hash function for each column value in a selected column of a data store table and may select a first domain and a second domain of hash values. The processor may determine a frequency value for each hash value within the first domain, generate a unique identifier for each hash value within the second domain, and determine at least one statistic on the selected column based on the frequency values and the unique identifiers. The processor may store the at least one statistic for use in a query plan. A method and computer-readable medium may also be implemented.

Abstract: A database system may include a storage array that includes a plurality of storage devices configured to store a database. The database system may further include a processor in communication with the memory device. The processor may determine frequency of data values of a first set of data from the database. The frequency of data values are determined at a predetermined data granularity. The processor may also generate a data object to include information indicative of the frequency of data values. The processor may also store the data object in the storage array. A method and computer-readable medium may also be implemented.

Abstract: Techniques for transitioning between code-based and data-based execution forms (or models) are disclosed. The techniques can be used to improve the performance of computing systems by allowing the execution to transition from one of the execution models to another one of the execution models that may be more suitable for carrying out the execution or effective processing of information in a computing system or environment. The techniques also allow switching back to the previous execution model when that previous model is more suitable than the execution model currently being used. In other words, the techniques allow transitioning (or switching) back and forth between a data-based and code-based execution (or information processing) models.

Abstract: Data of a database (e.g., database tables) can be reassigned from a first map to a second map in a database system that uses maps to assign data for processing to multiple processing units of a database system in accordance with one or more distributions schemes. Data portions can be selected in groups and moved in the selected groups in an efficient manner. The selection and/or movement of the data portions can be automated without requiring input for users of database systems.

Abstract: Methods and an apparatus for data sorting is provided. Keys are derived from a data set and a mapping function is obtained for sorting the data set in accordance with the mapping function. A wide key sort on the keys is performed over a plurality of distributed nodes using the mapping function, resulting in sorted lists of rows from the data set produced in parallel from the nodes with each row associated with a unique one of the keys pushed to a stack machine. The sort process is an ordered row traversal from the stack machine.

Abstract: A database operation is performed in a file system residing on a plurality of processing modules. The file system includes a first relation having a plurality of first-relation entries. Each of the plurality of first-relation entries has a first-relation attribute that is of interest in the database operation. A value of a distribution attribute in each of the first-relation entries is set to a unique value selected from among a domain of unique values. The first-relation entries of the first relation are redistributed among the plurality of processing modules based on the first-relation attribute and the distribution attribute. The computational operation is performed to produce a result.