IP Questions

[vpalania]

1.What is the adv. of having IP checksum covers only the datagram header & not the data ?

2.What is the MTU size for FrameRelay ? ATm network ?

3.What are the advantages of reassembly at ultimate destination instead of doing it after one network ?

Thanks,

Vijay

[lgijssel]

1: It requires less CPU-overhead to determine the validity of an IP header when only a part of the datagram must be checked.

2: It depends, I have no short or absolute answer to this.

3: Keep in mind that IP is essentially a connectionless protocol. This means the intermediate nodes have no knowledge of anything beyond the IP header. Reassembling and perhaps again fragmenting the packets per hop on the way through the network would require more cpu cycles and more intelligence on all intermediate nodes. It makes therefore more sense that reassembly is done only once at the final destination.

Regards,

Leo

[pkhatri]

Hi Vijay,

1. Most data-link layers have checksums that cover the entire payload. Therefore, it would be redundant to also verify the IP header checksum at every hop. In fact, this fact was recognised when designing IPv6 and the IP checksum has been removed from IPv6.

2. Most devices will support a FR MTU of 1600 bytes. According to the q.922 standard, all FR frames have to support a minimum of 260 bytes. For ATM, the MTU used is generally 4470 bytes.

3. Fragmentation/re-assembly is a CPU-intensive task and is generally considered bad. If routers before the final destination re-assembled the packets, there is a chance they could be fragmented again at some point further downstream. That would be needless work so that's why it's only done at the receiver. Once again, IPV6 recognised these and forbids routers from fragmentating packets altogether.

Hope that helps - pls rate the post if it does.

Paresh