Server group binding policy

The distribution of server groups of a single service or application on a cluster depends on the tasks being solved:

• If the service has several groups of servers (VMs, DB instances) with the same role, for example, several application servers, distribute them across different physical servers to increase fault tolerance.
• If there are groups of servers with different roles in the service for which communication latency is critical (for example, an application server and a database server), place them on the same computing node so that traffic passes within the same physical server.

VK Cloud supports four server group allocation policies, for soft policies, support for the compute-api version above 2.15 is required.

To define the rules for distributing server groups, specify the group in the instance properties:

1. Output a list of server groups using the command:

   Copy
   openstack server group list

2. Find a group of servers with a policy like affinity or anti-affinity. If no server groups are found, create one using the command:

   Copy
   ​openstack server group create <name of the server group> --policy <policy name>

3. Create bootable disks for future VMs. Example for three disks:

   Copy
   for i in 1 2 3;do openstack volume create --size 10 --image 98af6254-XXXX-XXXX-XXXX-81858ce9302a --availability-zone MS1 --bootable root-volume-$i;done

4. Create the necessary virtual machines specifying the server group. Example for three VMs:

   Copy
   for i in 1 2 3;do openstack --insecure server create --flavor std1-1 --volume root-volume-$i --hint group=<politic group ID> --nic net-id=<network ID> vm-affinity-$i;done

5. Wait for the VM creation to finish. Make sure that the deployed VMs have been successfully created and are in the status ACTIVE.