We're investigating creating a system where we have a camera transmitting large format video over gigabit Ethernet. It is a 16MegaPixel camera with 4 streams each transmitting 1024x4096 pixels (1/4 of the image on each stream). Our intent on the receive side, is to have a beefy PC (or dedicated box) with a quad port NIC, probably with the Intel pro1000 chipset, receiving the data. We already have proven out this setup using a board outputting GigE data over a single port to a PC receiving and displaying the data using a single port Intel pro1000 NIC.
From some preliminary research, it seems the best method to implement a 4 stream GigE camera is to using Static Link Aggregation. I see various sources on the web talking about
Cisco's proprietary "Port Aggregation Protocol" being used when teaming Intel NIC's together in a SLA arrangement. My question, will our embedded system that transmits the 4 streams of camera data have to adhere to this protocol? Keep in mind, this is a point-to-point connection using crossover cables from our camera to the PC receiving the data. No switch will be in the mix.
Secondly, it'd be nice to have one, fixed port be dedicated as the command and control port. I see in certain documentation talk of assigning primary and secondary ports. Would I have the capability to assign one of the NIC's as the command and control channel (i.e. all packets outgoing from my receiving application go out a certain port)? I'm only in need of the teamed bandwidth increase on the Rx side of my receiving application.