Abstract: An approach is provided in which the approach segments each one of multiple components corresponding to multiple component levels in an SQL database system into multiple functions. The approach combines a first one of the multiple functions with a second one of the multiple functions into an image, and invokes the image to process an SQL query using the first function and the second function.

Abstract: Building a first layer model of a three-layer model based on attributes that are sensitive features is provided. A first dimension reduction of sensitive features removes each sensitive feature having an indicator that it is present in a patch and does not contribute to one or more of three probabilities. A second dimension reduction of insensitive features is performed using vectorizing and using one-hot encoding. The remaining insensitive features are main features. One or more second layer models of the three-layer model is built based on the main features. The third layer model is built based on a verification dataset and the first layer model. Regression test coverage is recommended based on prediction result of the third layer model, and wherein regression tests are selected. The training dataset is updated based on probability calculations of the first layer model.

Abstract: An approach for optimizing server application response times. The approach creates a trust sharing context between edge clients and a server application. The approach identifies similar requests from the edge clients to the server application. The approach integrates the similar requests into a single request and normalizes the single request into a normalized data structure. The approach sends the single request to the server application for processing and receives the server application response to the single request. The approach distributes at least a portion of the response to the edge clients.

Abstract: Embodiments of the present disclosure relate to a method, system, and computer program product for generating one or more in-memory data structures for data access. According to the method, target data associated with a database is identified. Further, the method determines at least one data structure for the target data based on at least one access pattern of the target data in a plurality of historical queries against the database, wherein the target data is accessed in execution of the plurality of historical queries. The method further implements the at least one data structure in a memory to store the target data. The at least one data structure is used for further access to the target data in execution of a further query against the database.

Abstract: An approach is provided for automatically updating a data type of a database column. Based on a database definition, category, and usage, the column is determined to be a candidate for a data type change. For first values in the column values, costs of processing queries using a new data type are determined to be less than other costs of processing the queries using other data types. For exception value(s) in the column values, costs of processing the queries using the original data type are determined to be less than other costs of processing the queries using other data types. The original data type is replaced with the new data type for the first values and the original data type is retained for the exception value(s). The first values are associated with a first schema and the exception value(s) are associated with a second schema.

Abstract: Building a first layer model of a three-layer model based on attributes that are sensitive features is provided. A first dimension reduction of sensitive features removes each sensitive feature having an indicator that it is present in a patch and does not contribute to one or more of three probabilities. A second dimension reduction of insensitive features is performed using vectorizing and using one-hot encoding. The remaining insensitive features are main features. One or more second layer models of the three-layer model is built based on the main features. The third layer model is built based on a verification dataset and the first layer model. Regression test coverage is recommended based on prediction result of the third layer model, and wherein regression tests are selected. The training dataset is updated based on probability calculations of the first layer model.

Abstract: A computer-implemented method provides application-based query transformations. The method includes determining an application is initiated. The method includes identifying a set of execution units included in the application. The execution units are based on of a set of queries in the application and a set of actions in the application. The method also includes building a query dependency graph (QDG) comprising a plurality of nodes, wherein each node of the plurality of nodes is correlated to an execution unit, and each node is linked to at least one additional node, the link indicating a relative execution order and a common attribute each node and the additional node. The method includes merging, based on a performance architecture, two or more of the set of execution units into a section. The method includes processing the application according to the QDG.

Abstract: A computer-implemented method dynamically detects and corrects an error in a query. The method includes identifying an error in a first query. The method further includes generating a set of alternate execution structures for the first query. The method includes running each of the alternate execution structures, including generating a set of results corresponding to each set of alternate execution structure, comparing each of the set of results against each other of the set of results, and storing each of the set of alternate execution structures to include a result of the set of results, for each alternate structure. The method further includes selecting, from the set of alternate execution structures, a first alternate execution structure based on a predetermined criteria, and implementing the first alternate structure in place of the first query.

Abstract: Processing range operation data management statements in a database is provided. The method comprises receiving statements for range operations that specify referenced pages in the database. The range operations are stored in a search structure in a table directory in the database and applied to any referenced pages in a memory buffer pool. Application of the range operations is postponed for any referenced pages not in the memory buffer pool. The database determines if reading the postponed pages into the buffer pool would exceed a specified input/output threshold. If reading the postponed pages into the buffer pool does not exceed the specified threshold, the database reads the postponed pages from disk to the buffer pool asynchronously in parallel, and the range operations are then applied to the postponed pages. Pages modified by the range operations are then written from the buffer pool back to disk.

Abstract: Determining which data should be located in each partition of a hard disk drive storage device based on growth values and/or data access frequency values.

Abstract: A Generative Adversarial Network is used to train and/or tune a model used to analyze data in a database or data stream. The Generative Adversarial Network intermittently trains or tunes the model as the database is actively ingesting data and/or while the data stream is streaming. This intermittent refreshing of the model, performed by the Generative Adversarial Network, is sometimes described as “dynamic” or “dynamical.” Analytics type software is queried in order to perform normalization and/or model training.

Abstract: Increasing index search efficiency and availability in a database is provided. An index space matrix corresponding to a plurality of mini-indices of the database is built using ordered boundary values that correspond to key value ranges of each respective mini-index to increase index search efficiency by removing redundant mini-index accesses. Mini-indices of the plurality of mini-indices are consolidated asynchronously without suspending access to the mini-indices using a particular consolidation policy of a plurality of consolidation policies that corresponds to an amount of key value overlap identified between the mini-indices. Data operations corresponding to transactions are performed using the index space matrix during consolidation of the mini-indices without suspending access to the mini-indices.

Abstract: A computer-implemented method improves the performance of write ahead logging. The method includes generating a set of query data and a set of log data, where the set of log data is configured to create a write ahead log, and the set of query data is configured to make changes to data in a database. The method also includes writing the set of query data to a virtual file system. The method further includes separating the set of log data into a set of control intervals, where each control interval includes an entry. The method includes writing, each entry into a log buffer, where the writing the set of query data to the virtual file system and the writing each entry into the log buffer are performed in parallel. The method also includes combining each entry into a complete log.

Abstract: A processor may allocate a first buffer segment from a buffer pool. The first buffer segment may be configured with a first contiguous range of memory for a first data partition of a data table. The first data partition comprising a first plurality of data blocks. A processor may store the first plurality of data blocks in order into the first buffer segment. A processor may retrieve the target data block from the first buffer segment in response to a data access request for a target data block of the first plurality of data blocks.

Abstract: An approach is provided for optimizing a system resource of a cloud database. Components of a database system are divided into micro-systems according to functions and execution levels. A cluster analysis of the micro-systems and an analysis of workload patterns are performed. Different combinations of the micro-systems are generated. Images of the micro-systems and of the different combinations of the micro-systems are generated. A query is received and analyzed at a current layer specifying a set of micro-systems specified by a function of the database system. Service(s) associated with micro-system(s) specified by next layer(s) are pre-loaded and activated. A partial execution of the query is performed and a result of the query is generated at a selected edge or client side, where the selection is based on the analysis of the workload patterns.

Abstract: The invention provides a federated model based on locally trained machine learning models.

Abstract: A method includes: obtaining, by a database server, security policies associated with a database hosted by the database server; obtaining, by the database server, first database data associated with the database; creating, by the database server, a knowledge distillation model using the security policies and the first database data; obtaining, by the database server, second database data associated with the database; creating, by the database server, a classification model using the second database data and an output of the knowledge distillation model; receiving, by the database server, a client database request to the database; determining, by the database server, a new security policy rule set using the classification model; and applying, by the database server, the new security policy rule set to the client database request.

Abstract: An approach is provided in which a method, system, and program product build a time series prediction model based on one or more relationships between a first set of keywords in a set of first news articles and a second set of keywords in a set of second news articles. The time series prediction model includes a time-based interest level adjustment corresponding to a publication time between the set of first news articles and the set second of news articles. The method, system, and program product use the time series prediction model to compute an inherited initial interest level of a third news article that includes a set of new keywords based on the set of new keywords and the time-based interest level adjustment. The method, system, and program product assign the inherited initial interest level to the third news article.

Abstract: Methods, computer program products, and systems are presented. The method computer program products, and systems can include, for instance: receiving an incoming query statement, wherein the incoming query statement comprises a query statement expression that includes an input variable; predicting an input variable value associated to the input variable; selecting an access path for runtime execution of the query statement in dependence on the predicted input variable value; and performing runtime execution of the query statement using the selected access path.

Abstract: In an approach, a processor obtains a plurality of security policies for managing security of a database, the security policies defining respective sets of items to be monitored. A processor stores, respectively, the sets of items defined by the plurality of security policies in one or more item tables based on a predetermined criterion, the predetermined criterion comprising security policy relevance, security policy structure, and available space in the one or more item tables. A processor creates, for each of the one or more item tables, one or more respective index tables associated with the items in the respective item table. A processor creates, for each of the one or more item tables, a fast traverse block (FTB) associated with the created one or more respective index tables.