Very confusing explanation for someone that does not know the ways of a distributed system.
Multiple SpectroSERVERs, 1 landscape = Fault Tolerant installation.
One SpectroSERVER starting with a precedence of 10 (default), the second one with precedence 20. The lowest level precedence running SpectorSERVER is the active one at a given time. So basically you have the same devices in 2 servers. One is an active backup.
Multiple SpectroSERVERs, multiple landscapes (2+) = Distributed Installation.
Each SpectroSERVER is monitoring different devices. What my predecessor is describing is called Fault Isolation in Distributed Installation. To benefit from system advanced analysis algorithms you have to control how devices are connected in the distributed installation. If there are any links between devices that reside in 2 different landscapes, then you should use what's called a Proxy model. Basically it's a device that resides in 2 different landscapes, but it's only creating alarms in one landscape. It is participating however in Fault Isolation, so you can understand what's the root cause of a problem at a given time.
There's also the third option: Distributed Fault Tolerant installation. Each SpectroSERVER in the Distributed installation has at least one FT SpectroSERVER running to takeover whenever the primary fails.
So, to distribute the load across multiple servers, you need a Distributed Installation, meaning more than one landscape.
You need to split the devices on two SpectroSERVERs. The Distributed installation is fairly simple: you just install the second SpectroSERVER with a new landscape handle and point the Location Server to the existing SpectroSERVER.
You have to move your devices from the exiting landscapes to the new one. Nowadays this is performed using Modeling Gateway.