* Some info may not apply to Single Node Systems.
Scale Computing recommends these guidelines to keep a system running SC//HyperCore™ HyperVisor stable and redundant at all times:
- Leave a node’s worth of RAM free at all times so there is room for automatic failover. For example, in a four-node SC//HyperCore system, the HUD should show roughly no more than 75% of RAM in use.
- Put CPU-intensive VMs on separate nodes. Running several CPU-intensive VMs on one node may lead to VM performance issues.
- Use a fresh installation of a manufacturer-supported OS on a VM. This gives the most stability, especially if a migration would bring over an OS that the manufacturer no longer supports. See Supported Guest Operating Systems for tested guest OSes and minimum SC//HyperCore versions.
- Make sure the licensing you have for the OS is approved for virtualization of the guest VM, and for failover between nodes.