One question about RFC 1042

[sheepuking wu]

Hello,

Recently I study RFC 1042 and I have one question about this document.

I find that RFC 1042 has some points as follow:

1. It described the packet format that encapsulates IP datagram and ARP into L2 frame.

2. It described tha maxium/minum packet size of IP datagram and ARP packet.

But RFC 1010 has described the packet format that encapsulates IP datagram and ARP into L2 frame, and maxium/minum packet size of IP datagram and ARP packet can be computed if we know MTU of IEEE 802.3 or 802.4 or 802.5. I think every concept described in RFC 1042 had described in some past documents (ex, RFC 1010, IEEE 802.2, 802.5.....etc  ). So I am confused: what is motive that these authors want to write RFC 1042 ?

Thank you.

[lgijssel · Accepted solution]

Hi,

RFC1042 describes the use of the 802.2 and SNAP headers to provide a uniform method of encapsulating IP, regardless of the underlying media. In fact, RFC1042 is pretty ancient itself and the SNAP encapsulation only rarely used these days.

Encapsulation on Ethernet is now generally using the Ethernet II method.

RFC1010 is actually one of the first "assigned numbers" RFC's, an attempt to create an overview of all that was being assigned to a specific protocol or vendor. This is not to be confused with the more specific topic in RFC1042.

regards,

Leo

[Peter Paluch · Accepted solution]

Hello,

My personal opinion is that this RFC was written in times when it was not entirely clear what will be the predominant Ethernet frame format. The sleek Ethernet II (also called DIX) format ultimately won because of its simplicity and generally low overhead but in 1988 and even after, there were many networks using 802.2 LLC or SNAP encapsulation. Novell in particular was using 802.2 LLC encapsulation with its IPX/SPX suite, AppleTalk and DECnet ran on top of SNAP frames if my information is correct.

So it probably seemed to be a wise precaution to precisely specify how should the IP+ARP be encapsulated into IEEE 802 framing structures, even though these frame format are not used today to carry IP traffic.

Best regards,

Peter

[sheepuking wu]

Hello Peter and Leo,

Thanks for your help.

But RFC 1010 had described the SNAP org code = 0 and ehthertype = 0800 or 0806 if IP datagram or ARP is encapsulated into IEEE 802, so I think the concept about how should the IP+ARP be encapsulated into IEEE 802 was exist before RFC 1042 was released .

So now I have a problem: Since the concept about how should the IP+ARP be encapsulated into IEEE 802 was specified before, why  these authors want to write RFC 1042 to describe this concept ? 

Thank you.

Best regards,

Wu.