High Availability for VMs
Turbine includes powerful High Availability functions to keep your VMs going if hardware fails.
Clustering and High Availability
VM Squared’s robust clustering and host detection ensure that VM Squared doesn’t only detect when new hosts appear in the cluster. VM Squared also knows when hosts are experiencing problems and are no longer available to run VMs.
Hardware failures can and will happen. Applications continue to run thanks to VM Squared and integrated High Availability.
Turbine’s scheduler automatically reschedules VMs onto available hosts if a host is down. Turbine triggers actions when a host reaches an ERROR state to migrate the VMs to a running host.
Resources for High Availabiliy
As an administrator all you need to do is ensure that sufficient RAM and CPU capacity are available to cope with High Availability incidents.
For example, there is a cluster of 10 hosts and all VMs must remain on with the loss of two hosts. Each host has 128 GB; therefore, 10 hosts have a total RAM capacity of 1280 GB.
The usable RAM capacity required to maintain all VMs running when two hosts are lost can be calculated as: 1280 GB multiplied by the ratio of minimum remaining hosts to a full cluster size or,
This illustrates that the usable RAM in this scenario is 1024 GB, which allows for all VMs to continue running after two hosts fail. If the overcommitment of resources is in use, then simply substitute the overcommitted values for the TotalCapacity. The same equation is true for physical CPUs.
Neutron shared storage ensures that VM disks are accessible from any host in the cluster. VM Squared automatically configures Neutron and Turbine which means everything is ready to go from the start. VMs can be instantiated from any host in the cluster, as Neutron storage is available throughout the cluster. Even hosts which are not contributing disks to Neutron can still access disk images.
To only run VMs on specific hosts in the cluster, please add rules as shown in Scheduling Virtual Machines.