The limits of a Horizon Connection Server are defined by the maximum number of simultaneous client connections it can support. In Horizon 7, if you create a locked.properties file as described in the Horizon 7 Architecture Planning guide (https://docs.vmware.com/en/VMware-Horizon-7/7.6/horizon-architecture-planning.pdf), the maximum number of connections supported is 4,000 per Connection Server. If you do not create that file, the maximum number of connections supported is 2,000 per Connection Server.
Each single Horizon installation supports a maximum of seven Horizon Connection Servers in a 5+2 configuration known as a Pod, supporting a maximum of 10,000 active sessions. The term Pod implies that the Horizon environment is managed as one entity, based on the limitations of the Connection Server. This means that five of the servers are considered active, and two are held in reserve as spares. The spare Horizon Connection Servers maintain an up-to-date copy of the Horizon AD LDS database so that they can serve clients immediately in the event they are needed.
Not every Horizon environment will require seven Horizon Connection Servers, but at the very least they should have two. Even if the number of desktops to be deployed is less than the concurrent connection limit of 4,000 clients per Connection Server, having two Connection Servers ensures that sufficient capacity exists in the event of a server outage or maintenance event. Always build in extra capacity to accommodate events that can impact availability.
In environments where more than 10,000 virtual desktops are required, you will be required to deploy a second Standard Connection Server, which creates another Horizon Pod. This will create a unique Horizon AD LDS database, rather than a replica of the one in your first Horizon installation, enabling support for additional Horizon Connection Servers and an additional 10,000 desktops. Using the Horizon Cloud Pod feature described in Chapter 6, Implementing a Horizon Cloud Pod, you can link up to Pods together, and create a global entitlement that grants simultaneous access to Horizon pools in each.
A single-Pod design is suitable for smaller single-site Horizon environments as it is the easiest to manage. If you plan to install Horizon Connection Servers in multiple sites, you are required to create a unique Pod in each site as VMware does not support a single Pod that spans multiple sites due to the amount of communication traffic generated by connection replication.
In Chapter 6, Implementing a Horizon Cloud Pod, we will create a Cloud Pod, which is a collection of up to 25 Horizon Pods located at a maximum of ten distinct sites. While the Pods are still administered individually, using the Cloud Pod feature we can entitle users to Horizon resources located in multiple Pods from a single interface, and regardless of which Pod a user logs into they will be transparently connected to the appropriate Pod where their Horizon resources are located.