The bus is essentially the channel the data flows through to and from your hard drive. You can think of it as the bottleneck of a bottle. For the liquid to get from the inside of the bottle to the outside world it has to travel through the bottleneck (which tends to limit the flow of liquid). In the case of (modern) internal hard drives, the bus is SATA. The bottleneck is the bandwidth of SATA. For external hard its a variety of interfaces such as USB 2.0, 1394 (firewire), ethernet, wireless, bluetooth, esata, etc. The bottleneck in each one of these cases is essentially the bandwidth of the interface its connected to. None of these external interfaces are as fast as SATA (except for eSata which uses the SATA bus).
The reason rpms (on mechanical hard drives) are important is because the datarate of the harddrive have not saturated (i.e used up) the bandwidth of SATA (either 150 MB/s or 300 MB/s). So, in that bottle analogy, its like we are pouring liquid out of a glass instead of a bottle; water's flow out of the glass is not limited at all by the opening (top/rim) of the glass. So RPM's allow us to get faster and faster until at some point, our glass turns into a bottle (with a bottleneck). Essentially though, mechanical hard drives will never get fast enough to saturate SATA 2 as we are moving on to SSD's. SSD's are pushing close to the 3 GB/s boundary already. But once they get closer, we'll just make an even faster SATA specification (or a different specification if this becomes infeasible).
Regardless, nothing I am saying now will be that important in the future. A version of USB 3.0 has already been certified which supports 5 GB/s (only Linux supports USB 3 now even though afaik there are no USB3 supported motherboards). A newer version of firewire (not sure of the release) will support 6.4 GB/s. There is a newer version of SATA which supports 6 GB/s. As you can see, the newer firewire will be faster than the newer SATA. However, unless a hard drive is made with native firewire support (which won't happen) you will always have "bridging inefficiency" translating the meaning of the firewire ports to the SATA interface. Essentially, internal SATA will still be faster than a translated SATA on a faster bus. However, the new firewire would probably be a dagger into the eSATA market.