I am guessing you saw an old fashioned KVM that uses a standard KVM cable which has a "Keyboard, Video, and Mouse" port for each machine plugged into the KVM.
Those are quite rare, nowadays.
And then you saw a fancy new Avocent KVM, which has a RJ45 port for each server, and a SIP or Interface "POD" that each server plugs keyboard, video, and mouse into,
to carry the signal digitally over a single RJ45 cable?
This is what most datacenter KVM switches look like nowadays, whether made by HP, Dell, IBM, etc.
I don't believe i've seen a KVM switch with keyboard, video, or mouse ports on it in a very long time.
It's really the same thing as the older KVMs, it's just that a different type of media is used to carry the signal.
And the "PODs" the servers plug into are intelligent modules that transmit video signal to the KVM and inputs from the KVM, using a proprietary protocol.
By using PODs, standard RJ45 cabling can be used to make the run from the server rack to the KVM, there are fewer cables to deal with, it is tidier and more convenient.
But most importantly, it allows the KVM to achieve higher port densities, for example 48 RJ45 ports on a KVM takes up a whole lot less space than 48 monitor ports, 48 keyboard, and 48 mouse ports.
Is there something specific you would like to know about KVMs?
The newer KVMs are more flexible, because there are different type of PODs available.
For example, on the Avocent KVMs there are PODs for Video and PS/2 Keyboard and Mouse.
There are PODs for Video and USB Keyboard/Mouse
There are PODs for Video and USB 2.0
This way, different servers have their different interface needs accomdated, without requiring additional physical ports on the KVM itself.
For example, some servers might use only USB keyboards. Some server equipment may not support USB and support only PS/2.
There are even PODs for "serial" / console connections, useful for router/switch console access, essentially theoe are Terminal emulators,
you plug into the KVM, which attaches to a serial port instead of video and mouse. Where the "video" generated is virtual video provided by the POD,
and the "keyboard" input is translated into messages to a serial console.
This is useful (kind of)
