Keyboard, video and mouse signals carried over an IP network instead of dedicated KVM cabling, letting any operator desk reach any source PC.
What it is
IP-KVM extends classic KVM (see KVM) over a standard IP network. An encoder sits at the source PC and captures its display, USB and audio. A decoder at the operator desk reconstructs them in real time and forwards the operator's inputs back. Modern systems run encoder / decoder as software on commodity hardware or as zero-client appliances.
Why it replaced matrix KVM in control rooms
- Any-to-any reach — an IP network beats a fixed matrix at scale, because every operator desk can reach every source PC, including PCs in other rooms or cities.
- Same fabric as video sources — fewer cable types, fewer infrastructure silos.
- Software composability — an IP-KVM stream can be a source on a video wall, not just a desk-side session.
The hard parts
- Latency budget — operator input round-trip must stay under ~50 ms or "feel" breaks down. Achievable on tuned 1 GbE; trivial on 10 GbE.
- Security — every KVM stream is a remote-desktop attack surface. Look for AES encryption at the transport layer plus user authentication in the control plane.
- Failover — losing a decoder mid-incident is a real outage. Stand-by decoder + fast switchover is the usual mitigation.
Software-only IP-KVM
Modern video wall platforms (Craft Wall included) implement IP-KVM in software: encoders run on existing source PCs, decoders run inside the wall controller, and the operator's browser is the control plane. This collapses the hardware matrix entirely. Compare with the older Adder / IHSE / G&D matrix approach in KVM.