CLOUD WORKLOAD MANAGEMENT WITH AUTOMATED WORKLOAD BIDDING
A computer-implemented method for workload management in a computer system is provided. According to the method, a first compute node broadcasts a workload bid request to a plurality of compute nodes, wherein the workload bid request includes workload parameters characterizing the workload. The plurality of compute nodes each receive the workload bid request from the first compute node, and each of the plurality of compute nodes uses the workload parameters included in the workload bid request to calculate a cost of running the workload. One or more individual compute nodes within the plurality of compute nodes each send a workload bid to the first compute node, wherein each workload bid includes the cost of running the workload on the individual compute node sending the workload bid. The first compute node receives each workload bid and selects a target compute node to run the workload.
1 . A computer-implemented method for workload management, comprising:
a first compute node broadcasting a workload bid request to a plurality of compute nodes, wherein the workload bid request includes workload parameters characterizing the workload;
the plurality of compute nodes each receiving the workload bid request from the first compute node, wherein each of the plurality of compute nodes uses the workload parameters included in the workload bid request to calculate a cost of running the workload;
one or more individual compute nodes within the plurality of compute nodes each sending a workload bid to the first compute node, wherein each workload bid includes the cost of running the workload on the individual compute node sending the workload bid; and
the first compute node receiving each workload bid and using each workload bid to select a target compute node to run the workload, wherein the target compute node is selected from the one or more individual compute nodes and the first compute node.
2 . The computer-implemented method of claim 1 , wherein the first compute node and the plurality of compute nodes are included in a network.
3 . The computer-implemented method of claim 1 , further comprising:
the first compute node selecting to run the workload on the first compute node.
4 . The computer-implemented method of claim 1 , wherein the workload bid request includes a cost function that the plurality of compute nodes use to calculate the cost of running the workload.
5 . The computer-implemented method of claim 4 , wherein the cost function includes one or more factors selected from memory available, battery life remaining, time to process, and monetary cost.
6 . The computer-implemented method of claim 4 , wherein the cost function includes an optimization parameter selected from time, energy use, and monetary cost.
7 . The computer-implemented method of claim 1 , further comprising:
the first compute node broadcasting that the workload bid request has been awarded.
8 . The computer-implemented method of claim 1 , further comprising:
the first compute node broadcasting the cost included in the workload bid of the target compute node selected to receive the workload.
9 . The computer-implemented method of claim 8 , further comprising:
the plurality of compute nodes, other than the target compute node, receiving the cost included in the workload bid of the target compute node and tracking the cost and workload parameters of selected compute node for use in preparing workload bids for subsequent workload bid requests.