Simplifying provisioning of large homogeneous clusters

Current cluster create call requires information like flavor, volume size, neutron network etc to be specified on the instance level. Although it provides flexibility for customizing instances in a cluster, for larger homogeneous clusters where instance properties won't change very often it requires end user to pass in lot of redundant information.

If we can allow parameters like flavor, volume size, neutron network and availability zone on the cluster level as well as on the instance level, we will be able to avoid passing in the redundant parameters for larger homogeneous cluster and we will also be able to offer the flexibility to the user to customize the instances in their cluster if required.