I understand that ISP modem is a combo box made up of modem, router and switch. When it is in unbridged mode, the internal rotuer WAN simply recieve ISP ip address and its LAN interface is statically configured to be 192.168.1.1. This way the HOST can access modem configuration through 192.168.1.1.
Correct.
In your first diagram, the modem and switch don't have their own IP addresses; they're just chips managed by the same CPU which performs routing. (I...think a relevant term is "control plane"?)
(But a dedicated modem or switch device can and often will have an IP address for management.)
However when it is bridged the router component "dissapear". Conceptually it looks like [diagram]
Well, it's only removed from the packet path, but doesn't disappear entirely. It's still there, and it still has an IP address.
In router mode, an inbound Ethernet frame is processed like this:
- If the destination MAC matches the router's MAC, pass it up to the 'router', which will handle it according to destination IP.
- Otherwise, discard the frame.
In bridge mode, you could say that an inbound Ethernet frame is processed like this:
- If the destination MAC matches the router's MAC, pass it up to the 'router', which will handle it according to destination IP.
- Otherwise, forward the frame to other suitable bridge ports.
It's that 2nd part which lets your host bypass the router and talk directly to the ISP. But if your host tries to reach the modem/router/switch specifically, that still works exactly as before. (And depending on how it's configured, it could still act as a router too.)
(I wrote "could say" because the two steps actually get merged into one – the MAC is looked up in the bridge's FDB, and the frame either forwarded to an Ethernet port or to the "CPU" port. But for visualization purposes, the description is still acceptable.)
