Hello,
I am studying CEF and know the basic working of CEF. However, I need help in following questions:
Question 1: I understand that CEF uses mainly Prefix, NextHop, and Interface information in its table to forward packet. It seems that Packets are HW Accelerated based on only destination IP/Network. I was wondering if CISCO CEF supports forwarding of packets based on transport layer? For example, some other vendors provide this facility to HW accelerate the packets based on FLOW matching the source/destination port(s) and source/destination IP(s). For example, from other vendor, following entry takes src/dest ports into consideration in HW acceleration table to forward the packets.
Rule ID: 1
PROTO: 17
SOURCE PORT: 3333
DESTINATION PORT: 22
Source: 10.10.10.1
Destination: 20.20.20.1
GATEWAY: 10.10.10.2
Question 2: If the answer to the question 1 is YES, then i would also like to know how would CEF/Hardware Acceleration forward Fragmented Packets. As far as I know, when fragmentation is done, only first packet contains protocol information (TCP/UDP). All other subsequent packets don't. If CEF considered TCP/UDP in its CEF table, how would it process all subsequent packets that have no information of upper layer protocol (src/dest ports).
Question 3: This is separate question from CEF. If a bigger sized packet is fragmented into 4 fragments, is it possible to send 2 packets on Link-1 and other 2 packets on Link-2. In this way, all the intermediary routers will not reassemble the fragments. And the re-assembly only happens at the destination. Is this true ?
Also please, if possible, share when it is required to re-assemble fragments before even reaching to the destination. I have heard that when firewall is in between the Source and Destination, Firewall must re-assemble all the fragments. Is this true ?
Any help would highly be appreciated.
Thanks.