International Business Machines Corporation

Xiao Ming Bao

Yan Li

Tao Liu

Qi Guo

Zhenhua Hu

Wen Wei Tang

Jian Qiang Liu

Kuan Feng

Guan Cheng Chen

Wen Yan Bai

Haohai Ma

Khalid Ahmed

Junfeng Liu

Su Lei

Jun Feng Liu

In at least one embodiment of stream processing in a data processing system cluster, a driver requests, from a cluster manager, execution of a first task within a job including a plurality of tasks, as well as a first quantity of resources of the cluster to support execution of the task. At completion of execution of the first task by the cluster, the driver retains for a second task among the plurality of tasks at least a portion of the resources allocated to the first task by the cluster manager. The driver determines a second quantity of the resources to retain for the second task based on a relative priority of the job to other jobs in the cluster and efficiency of the cluster manager in allocating resources to the first task.

Adaptive resource scheduling for data stream processing

Embodiments for graphical processing unit (GPU) resource sharing in a computing cluster, by a processor device. Resource-specific stages are dynamically generated in a directed acyclic graph (DAG) using a DAG interpreter for a set of tasks by creating equivalence stages in the DAG having an associated inserted set of shuffle stages, the equivalence stages created based on a determined cost of each stage of the set of shuffle stages. Backlog tasks are scheduled and tasks within the set of tasks are shifted among respective stages in the equivalence stages according to the determined cost to avoid overlapping allocation of the GPU resource during central processing unit (CPU) execution of the respective tasks of the set of tasks.

Cluster graphical processing unit (GPU) resource sharing efficiency by directed acyclic graph (DAG) generation

In an approach to adaptively pipeline a MapReduce job, a processor receives one or more data records from a storage and inserts the one or more data records into a map queue, wherein a size of the map queue is adaptive to one or more utilizations of one or more resources in the processor. One or more processors apply a map function to the one or more data records in the first buffer and sort the records that are output from the map function and store the sorted records. One or more processors receive and insert the sorted records into a reduce queue, wherein a size of the reduce queue is adaptive to one or more utilizations of resources in the one or more processors. One or more processors apply a reduce function to the sorted records in the reduce queue and store a result in a storage.

Adaptive Map-Reduce pipeline with dynamic thread allocations

Adaptive map-reduce pipeline with dynamic thread allocations

The present invention relates to a method, apparatus, and computer program product for scheduling in job execution. According to embodiments of the present invention, there is provided a method for scheduling a plurality of job slots shared by one or more pre-processors and one or more post-processors in job execution, wherein the data generated by the pre-processor(s) will be fed to the post-processor(s) for processing. The method comprises: determining an overall data generation speed of the pre-processor(s); determining an overall data consumption speed of the post-processor(s); and scheduling allocation of at least one of the job slots between the pre-processor(s) and the post-processor(s) based on the overall data generation speed and the overall data consumption speed. Corresponding apparatus is disclosed as well.

Scheduling in job execution

Growing community of inventors

Beijing, China

Xiao Ming Bao