Oracle International Corporation

Oracel International Corporation

Arvind Kumar Maheshwari

Raghav Ravichandran

Kusumaharanadh Poduri

Vamshidhar Reddy Pasham

Ganesh Narayanamurthy

Rishi Saraswat

Karl Dias

Uri Shaft

Vladimir Volchegursky

Vishwanath Karra

Tse-Han Huang

Stephen Wexler

Shantanu Mahajan

Debottam Kundu

Anil Kumar Kothuri

Ravi Kumar Agrawal

Techniques are described herein for generating, editing, and optimizing queries using neural networks. In some embodiments, the techniques include training a neural network using a set of performant database queries to automatically learn patterns between different sequences of tokens in performant queries. Once trained, the neural network may receive an incomplete query as input, where the incomplete query includes one or more query tokens. The trained neural network may then perform next token prediction to project a set of one or more additional query tokens that may follow the one or more query tokens in the incomplete query to form a completed, performant query.

Intelligent query editor using neural network based machine learning

Techniques are described herein for leveraging recurrent neural networks for query processing. In some embodiments, a query analytic system determines a sequence of tokens for at least a portion of a query and determines a vector representation for each token. The query analytic system further generates, using a neural network based on the sequence of tokens, a performance prediction associated with executing at least the portion of the query, wherein the neural network assigns at least a first weight for at least a first token in the sequence of tokens based at least in part on at least a second token that preceded the token in the sequence. The query analytic system further triggers a responsive action, such as triggering an alert and/or tuning the query, based at least in part on the performance prediction.

Query classification and processing using neural network based machine learning

Techniques are described herein for seasonal pattern determination and validation. In one or more embodiments, a set of time-series data is received to analyze for seasonal behavior. In response a plurality of patterns may be generated, including a first pattern and a second pattern, such that each of the first pattern and the second pattern approximate data points that represent a same sub-period of multiple instances of a season within the set of time-series data. One or more other instances of the season may then be analyzed to determine whether at least part of the first pattern or the second pattern is detected. Based at least in part on determining that the at least part of the first pattern is detected in the at least part of the same sub-period, a responsive action that is associated with the first pattern may be performed.

Seasonality validation and determination of patterns

Techniques for analyzing an execution of a query statement based on a random archive are disclosed. A plurality of query statements that are executed during a particular time period are identified. A random sampling function is executed to randomly select a set of query statements from the plurality of query statements. Execution plans and/or performance metrics associated with each execution of the randomly-selected query statements are stored into a random archive. Responsive to determining that a performance metric for a current execution of a particular query statement does not satisfy a performance criteria, information associated with the particular query statement from the random archive is analyzed. A model plan characteristic associated with an execution of the particular query statement stored in the random archive is determined. An execution plan associated with the model plan characteristic is determined for another execution of the particular query statement.

Database performance analysis based on a random archive

Techniques for selecting an anomaly based on a context are disclosed. A set of metrics corresponding to communications with nodes of a computer system are identified. A set of insights are generated based on the set of metrics. A context for determining a primary anomaly is determined. A subset of metrics associated with the context are identified. A subset of insights that are generated based on the subset of metrics are identified. An insight is selected from the subset of insights as the primary anomaly. A visualization associated with the primary anomaly is presented at a user interface. One or more secondary anomalies may be concurrently presented with the visualization. Additionally, the primary anomaly, the selected visualization, and/or the secondary anomaly is used to determine a new context for selecting another primary anomaly. Hence, a series of primary anomalies may be selected, each primary anomaly being related to each other.

Selecting an anomaly for presentation at a user interface based on a context

Methods for obtaining and analyzing a reduced data set are disclosed. In an embodiment, runtime dependencies between components are identified. Based on runtime dependencies, it is determined that a first component is configured to access a second component during runtime but is not configured to access a third component during the runtime. A subset of monitoring data associated with the second component is reported. A subset of the monitoring data associated with the third component is not reported. In an embodiment, a first data set associated with a component is obtained. The first data set is aggregated by monitoring of the component. A second data set, associated with the component, is simulated. The second data set is simulated based on the first data set and/or data modeling associated with the component. A characteristic and/or performance of the component is analyzed based on the first data set and the second data set.

Obtaining and analyzing a reduced metric data set

Various embodiments are presented for bulk recovery or faults in a service oriented architecture system. The number of faults submitted for recovery is determined based on the capacity of the system. A linear programming model is used to determine the maximum recovery capacity of the system. The maximum recovery capacity is configured to be below the capacity of the system.

Growing community of inventors

Pleasanton, CA, United States of America

Arvind Kumar Maheshwari