International Business Machines Corporation

Eric Allen Jacobson

Xiaoyan Pu

Yong Li

Krishna Kishore Bonagiri

Lawrence A Greene

Shyam R Mudambi

Kalyanji Chintakayala

Ritesh Kumar Gupta

Scott Louis Brokaw

Brian K Caufield

Namit Kabra

Jo Arao Ramos

Ron E Liu

Dong Jie Wei

Likhitha Maddirala

Indrani Ghatare

A method, computer system, and a computer program product for data pipeline prioritization is provided. Embodiments may include receiving, by a cognitive rules engine, one or more data pipelines. Embodiments may then include analyzing, using a computational method of the cognitive rules engine, the one or more data pipelines. Embodiments may lastly include prioritizing the one or more data pipelines based on a result of the computational method of the cognitive rules engine.

Prioritization and prediction of jobs using cognitive rules engine

A self-learning computer-based system has access to multiple runtime modules that are each capable of performing a particular algorithm. Each runtime module implements the algorithm with different code or runs in a different runtime environment. The system responds to a request to run the algorithm by selecting the runtime module or runtime environment that the system predicts will provide the most desirable results based on parameters like accuracy, performance, cost, resource-efficiency, or policy compliance. The system learns how to make such predictions through training sessions conducted by a machine-learning component. This training teaches the system that previous module selections produced certain types of results in the presence of certain conditions. After determining whether similar conditions currently exist, the system uses rules inferred from the training sessions to select the runtime module most likely to produce desired results.

Self-learning selection of information-analysis runtimes

A method includes: receiving, by a computer device, resource request for a data integration job, wherein the resource request is received from a job executor module and defines processes of the data integration job; allocating, by the computer device, containers for the processes of the data integration job; launching, by the computer device, a respective wrapper script on each respective one of the containers after allocating the respective one of the containers; and transmitting, by the computer device and in response to the allocating, node details to the job executor module. In embodiments, the wrapper script running on the container is configured to repeatedly check a predefined location for process commands from a job executor. After the resource manager allocates all the containers for a data integration job according to a resource request, the job executor writes the process commands to the predefined location. Each wrapper script continues to check the predefined location for the process command that it is assigned to run, and runs the process command as soon as it is available at the predefined location. The process commands may be indexed with index values matching those assigned to respective ones of the wrapper scripts.

Handling expiration of resources allocated by a resource manager running a data integration job

A set of resources required to process a data integration job is determined. In response to determining that the set of resources is not available, queue occupation, for each queue in the computing environment, is predicted. Queue occupation is a workload of queue resources for a future time based on a previous workload. A best queue is selected based on the predicted queue occupation. The best queue is the queue or queues in the computing environment available to be assigned to process the data integration job without preemption. The data integration job is processed using the best queue. It is determined whether a preemption event occurred causing the removal of resources from the best queue. A checkpoint is created in response to determining that a preemption event occurred. The checkpoint indicates the last successful operation completed and provides a point where processing can resume when resources become available.

Resolving container preemption

Disclosed aspects relate to data integration application execution management using both a set of managed nodes and a set of unmanaged nodes. An execution request for a data integration application may be sensed. A first set of commands of the data integration application may be assigned to the set of managed nodes. A second set of commands of the data integration application may be assigned to the set of unmanaged nodes. Execution of the data integration application may be initiated using the set of managed nodes to process the first set of commands and the set of unmanaged nodes to process the second set of commands.

Data integration application execution management

Provided are a computer program product, system, and method for allocating physical nodes for processes in an execution plan. An execution plan is generated indicating a plurality of processes. A resource requirement is generated indicating requested physical nodes and an assignment of the processes to execute on the requested physical nodes. A determination is made from the resource requirement of a resource allocation of physical nodes for the requested physical nodes and the processes. The execution plan is updated to generate an updated execution plan indicating the physical nodes on which the processes will execute according to the received resource allocation.

Allocating physical nodes for processes in an execution plan

Performing an extract, transform, and load (ETL) process. Column data is received by a stage of the ETL process. The size of the received column data is ascertained by the stage. In response to determining that the size of the column data exceeds a predefined threshold, the stage saves the column data and creates a data locator associated with the column data. The created data locator advances through successive downstream stages of the ETL process as a replacement for the column data.

Shipping of data through ETL stages

Shipping of data though ETL stages

At least one application in a computing environment is executed and one or more performance metrics of the application are measured. The measured performance metrics are analyzed and an operational performance regression is detected. The detected operational performance regression is correlated with one or more recorded changes and the correlated changes are identified as a cause of the operational performance regression. The elements of the computing environment are alerted in accordance with the identified changes to adjust operational performance.

Detecting causes of performance regression to adjust data systems

Embodiments presented herein provide techniques for optimizing parallel data flows of a batch processing job using a profile of the processing job. An application retrieves a job profile for a processing job. The processing job has a plurality of processing stages specified in an execution profile. The job profile includes statistical data for at least one of the processing stages obtained during prior executions of the job. The application modifies properties of the execution profile based on the job profile to optimize the execution of the job. The application executes the processing job with the modified execution profile.

Growing community of inventors

Arlington, MA, United States of America

Eric Allen Jacobson