This section describes features of interest to administrators.
Managing VM memory
Configure Dynamic Cluster to manage the amount of memory for VMs.
Enabling CPU binding Dynamic Cluster supports CPU binding of virtual machines in VMware and KVM environments.
Hypervisors as physical machines
I/O operations are not efficient when executed on VMs, it is preferable to run I/O intensive jobs directly on physical machines. To maximize resource usage, Dynamic Cluster supports physical host repurposing between physical machines and hypervisors, based on job demand.
VM job preemption
Preemptive scheduling in LSF allows a pending high-priority job to preempt a running job of lower priority. LSF suspends the lower-priority job and resumes it as soon as possible.
Live migration script Dynamic Cluster supports the live migration of virtual machines with running jobs. The VMO agents running on both the source and target hypervisors will call out a script (live_migrate_script located in the /opt/platform/virtualization/4.1/scripts directory on the hypervisors) to perform the live migration.
Manual live migration of VM jobs LSF administrators may manually request a live migration of VM jobs from one hypervisor (the source hypervisor) to another hypervisor (the target hypervisor) by using the bmig command. The VM jobs remain running after the live migration. This is useful for removing running jobs from a hypervisor, for example, to prepare the hypervisor for maintenance.
Host memory defragmentation
Enable and configure host memory defragmentation to allow large memory jobs to run on large memory hosts.
Manually saving VMs to disk
VMs (and jobs running on them) may be saved to disk using the bmig command (normally used to migrate VMs).
Distinguishing provisioning time from running time Dynamic Cluster jobs will distinguish between provisioning time (when the jobs are in the PROV job state) and running time (when the jobs are in the RUN job state) to show a more accurate job status and be able to determine how much time jobs are spent actually running.
VM job checkpoint and restart
Checkpointing enables Dynamic Cluster users to pause and save the current state of memory and local disk of a VM running a job. The checkpoint files allow users to restart the VM job on the same physical server or a different physical server so that it continues processing from the point at which the checkpoint files were written.