Gary,
I'm sorry, but you are wrong. Any device that forwards a packet from one network port to another, looks at the packet. It has to, in order to perform a table lookup to determine what port the packet should go out.
Layer 2 switches only look at the physical layer address, i.e. ethernet's source MAC, to perform the table lookup. MAC addresses are always in the same place in the packet and of the same format, so it's easy to do this in hardware.
Layer 3 switches and traditional routers do the exact same thing. The only difference being they look beyond the physical address and use the network address, i.e the IP address, to perform the table lookup. This is still pretty easy to do in hardware, but IP options and various encapsulations complicate the process.
Where the complexity lies is in the code that actually performs the route determination, which then fills in the entry in the forwarding lookup table. Now we're talking about RIP, OSPF, BGP, IGP, etc. This is hard, very hard. Especially when you try to scale the software to handle thousands of routing entries and millions of packets per second.
Traditional routers and Layer 3 switches differ largely in what part of the system is handling the forwarding of packets. Traditional routers use software running on off the shelf processors to perform both route determination and forwarding. Very hard to do with high performance and reliability. Layer 3 switches tend to utilize custom chips, ASICs, to handle the forwarding of packets, i.e. once the route is known, and use traditional routing software on off the self processors to perform route determination.
Regards,
Lerxst |
