Adding to what Kyriakos has said... sprouting is more like something that turns your entire OpenFlow network into an L2 learning switch, so it behaves the same way -- if it doesn't know where a destination is, it floods the packet.
Why is this? Well, because there might be a path between the source and destination and it just doesn't know it yet. The only way it knows where anything is by seeing a packing FROM it and "learning" its position. So if host A tries to send to host B, but host B hasn't sent anything yet... you have no idea where host B is. The two may be connected; you just don't know it. You'd have to wait for host B to send something before you'd know where to direct the packets from host A. By flooding the packet from host A, host B will hopefully get the packet and then respond. Now you know where host B is and can install exact paths. -- Murphy On Jul 23, 2012, at 4:36 AM, 王健 wrote: > Hello guys: > "Default routing component. Listens for Flow_in_events and sets up > shortest path route from source to destination. If no route betweent the > access point datapaths exists, broadcasts packet." The sentence i saw in > src\nox\netapps\routing\sprouting.hh. My Question is why not drop this packet > if there is no route?
