System and method for managing a hybrid compute environment

摘要

Disclosed are systems, hybrid compute environments, methods and computer-readable media for dynamically provisioning nodes for a workload. In the hybrid compute environment, each node communicates with a first resource manager associated with the first operating system and a second resource manager associated with a second operating system. The method includes receiving an instruction to provision at least one node in the hybrid compute environment from the first operating system to the second operating system, after provisioning the second operating system, pooling at least one signal from the resource manager associated with the at least one node, processing at least one signal from the second resource manager associated with the at least one node and consuming resources associated with the at least one node having the second operating system provisioned thereon.

主权项

1. A method of managing workload in a hybrid multi-node compute environment in which at least one node has a first operating system and is controlled by a first resource manager and at least one node has a second operating system and is controlled by a second resource manager, the method comprising:
establishing a policy for the hybrid multi-node compute environment of balancing a proportion of nodes in the hybrid multi-node compute environment between the first operating system and the second operating system according to a certain parameter to yield an established policy, wherein the certain parameter comprises one of historical use, current activity, pending workload, current network environment, compute environment condition, anticipated workload, and history associated with workload submissions; receiving an instruction from a workload manager to provision a node in the hybrid multi-node compute environment from the first operating system to the second operating system; after provisioning the second operating system and as instructed by the workload manager: pooling, into a latency pool, a first signal from the first resource manager associated with the node and processing a second signal from the second resource manager associated with the node such that workload consumes resources associated with the node having the second operating system provisioned thereon; periodically analyzing the hybrid multi-node compute environment to determine how well it is aligned with the established policy to yield an analysis; and based on the analysis, dynamically provisioning nodes in the hybrid multi-node compute environment to change at least one node from the first operating system to the second operating system or at least one node from the second operating system to the first operating system according to the established policy.