The payload of the IP packet will depend on the frame that it is being encapsulated with. The IP packet can be fragmented (split into multiple frames), but this is generally undesirable, especially in cases of some encrypted payloads.
You can set the max with the MTU (maximum transmission unit)
Preamble is a series of alternating ones and zeros prepended to the frame. It's job is to give the receiving system something to synchronize with before the actual data in the frame.
Since Ethernet is asynchronous, there is no way for the receiver to predict when a frame might appear on the wire. When the preamble appears (as it must, by specification), the receiver uses the preamble bits to sync-up with the incoming frame so it will have a reference timing for the data that follows.
The preamble (64bits, I think) is also the reason for the 3-4-5 rule in traditional Ethernet.
A few of the preamble bits are lost/dropped/not seen when the frame appears at the receiver, so if the frame is repeated out (not regenerated), there are fewer preamble bits.
If the frame is repeated out enough times, then there are not enough preamble bits for the next receiver to synchronize with, and the frame is lost/discarded (frequently with CRC errors, since the receiver couldn't sync properly).
Good Luck
Scott
