Before installing the VMware Integrated OpenStack OVA, configure your environment as described in the following procedure.
Prerequisites
- Deploy vCenter Server and all ESXi hosts.
- Open the TCP and UDP ports required by VMware Integrated OpenStack. See Required Network Ports.
- Create a PTR record associating the IP address planned for the OpenStack Management Server with its FQDN, and ensure that the OpenStack Management Server can connect to a DNS server.
Procedure
- In vCenter Server, create a data center for VMware Integrated OpenStack.
- In the data center, create the management cluster and compute cluster.
- For HA deployments, the management cluster must contain at least three hosts and at least one datastore.
- For compact or tiny deployments, the management cluster must contain at least one host and at least one datastore.
- If you want to use NSX networking, create the edge cluster.
- Enable DRS and vSphere HA on the management cluster.
- If you want to use datastore clusters for compute nodes, enable Storage DRS on the datastore clusters and set the Cluster automation level to No Automation (Manual Mode).
- In your data center, create one or more distributed switches for your management, compute, and edge clusters.
- On each distributed switch, set the maximum transmission unit (MTU) to 1600 or higher.
- Plan the management network and assign a dedicated VLAN to it.
- Ensure that the management network has sufficient IP addresses to support your deployment. The required size of this network depends on the deployment mode and on whether you want to deploy Ceilometer.
- HA without Ceilometer: 11 contiguous IP addresses
- HA with Ceilometer: 16 contiguous IP addresses
- Compact or tiny without Ceilometer: 4 IP addresses
- Compact or tiny with Ceilometer: 9 IP addresses
- Ensure that the management network can be expanded to twice the original number of IP addresses during upgrades. When upgrading VMware Integrated OpenStack, you will temporarily require sufficient IP addresses to support two deployments.
- For NSX-T Data Center deployments, ensure that the vCenter Server, NSX Manager, and NSX Controller instances can access the management network on Layer 2 or Layer 3.
- For NSX Data Center for vSphere deployments, if you want to use the management network for metadata service, ensure that the management network has an additional two contiguous IP addresses available. This is not required for deployments with an independent metadata service network.
- Plan the API access network and assign a dedicated VLAN to it.
- Create the port group for the API access network on the distributed switches for your management, compute, and edge clusters. The network must be externally accessible.
- Ensure that the API access network has at least two contiguous IP addresses for HA deployments or at least one IP address for compact or tiny deployments.
- Ensure that the API access network can be expanded to twice the original number of IP addresses during upgrades. When upgrading VMware Integrated OpenStack, you will temporarily require sufficient IP addresses to support two deployments.
- For NSX Data Center for vSphere deployments, create the port group for the external network on the distributed switch for the edge cluster.
- (Optional) For NSX Data Center for vSphere deployments, plan the metadata service network.
- Ensure that the metadata service network has at least two contiguous IP addresses available. If you do not create an independent metadata service network, ensure that your management network has two additional IP addresses available for the metadata service.
- Ensure that the metadata service network can communicate with the management network.
What to do next
If you want to use NSX networking, deploy and configure your networking back end.
