There are some differences between the wizard screens when an ESG is deployed, compared to a DLR.
The first one is on the “Configure deployment” screen:
For an ESG, “Configure Deployment” allows selection of the Edge size. If an ESG is used only for routing, “Large” is a typical size that is suitable in most scenarios. Selecting a larger size will not provide more CPU resources to the ESG’s routing processes, and will not lead to more throughput.
It is also possible to create an ESG without deploying it, which still requires configuration of an Edge Appliance.
A “Non-deployed” Edge can be later deployed via an API call or with the “Deploy” UI action.
If Edge HA is selected, you must create at least one “Internal” interface, or HA will fail silently, leading to the “split-brain” scenario.
The NSX UI and API allow an operator to remove the last “Internal” interface, which will cause HA to silently fail.