Distribution of data and task instances in grid environments

A method comprising:

• determining a task instance to be executed;
• determining a data partition associated with execution of the task instance and stored at an initial node of a grid network;
• determining a processing node of the grid network having processing resources required to execute the task instance and a data node of the grid network having memory resources required to store the data partition, based on a latency therebetween;
• making a comparison as to whether an execution time of the task instance with the data partition at the data node will be less than a default execution time of the task instance with the data partition remaining at the initial node;
• deploying the task instance to the processing node;
• and
• re-locating the data partition from the initial node to the data node, based on the comparison,
• wherein making the comparison includes calculating the execution time as including the larger of a deployment time of deploying the task instance and a transfer time of transferring the data partition from the initial node to the data node.

A computer program product tangibly embodied on a non-transitory computer-readable storage medium and including executable code that, when executed, is configured to cause a data processing apparatus to:

• receive a task instance to be executed on a grid network, the task instance being instantiated from a task of an application;
• determine a data partition from a database associated with the application, based on the task instance, the data partition being associated with the task and stored on an initial node of the grid network;
• determine a processing node of the grid network having sufficient processing resources to execute the task instance;
• determine a data node of the grid network having sufficient memory resources to store the data partition;
• determine that an execution time of the task instance at the processing node with the data partition having been transferred from the initial node to the data node is less than a default execution time of the task instance at the processing node with the data partition remaining at the initial node;
• deploy the task instance to the processing node;
• and
• transfer the data partition from the initial node to the data node,
• wherein the execution time includes the larger of a deployment time of deploying the task instance and a transfer time of transferring the data partition from the initial node to the data node.