Cisco Support Community
cancel
Showing results for 
Search instead for 
Did you mean: 
Announcements

Welcome to Cisco Support Community. We would love to have your feedback.

For an introduction to the new site, click here. And see here for current known issues.

New Member

tcp-ip encapsulation

hello,

im a little confuse, i hope someone could explain me something -

the internet work with tcp-ip protocol -

now - all tcp-ip packet s are encapsulated and become a frame.

i know that ethernet have 4 type of encapsultaion for example -

802.2, 802.3, ethernet II

what happend to the tcp-ip packets in the second layer(data link layer),

in the internet.

------------

i know you can encapsulate tcp-ip with - ppp,frame relay, hdlc,

but my question is how the tcp-ip packets encapsulate in the second layer

in the internet.

is it 802.3 ?

i know that the arp protocol is getting the mac address - but how it encapsulate the packet with it ?

---------------------------------------------------------------------------------------------------------

other question i have -

what is the default second layer encapsultaion of tcp-ip in Lan

is it 802.2, 802.3 or what ?

can i determine what will be the encapsulation -

let say it 802.2 and i want it to be 802.3 can i do it ?

who is responsible for the second layer encapuslation -

is it the configuration of tcp-ip ?

as i said im a little confuse, i hope that someone could explain me

thank, --bye--

3 REPLIES
New Member

Re: tcp-ip encapsulation

To better understand how it works I refer you to Edward Comer, Internetworking with TCP/IP which is the comprehensive source for TCP/IP mechanics. Having said that let me see if I can shed some light on you question.

First of all you should have a little familiarity with the layered network model. In TCP/IP there are only 5 layers Physical/Datalink/Network/session/Application.

An application generates data and hands it over to TCP (or UDP) which encapsulates in a packet and hands it over to IP which encapsulate in a datagram and hands it over to datalink which encapsulates in a frame and hands it over to phisycal. Let's say we are in a LAN so you have an ethernet frame ...

Now you need to go on the Internet so the packets gets to the default gateway (which is a router) the frame is stripped and the data is handed to the network layer which makes a routing decision and forwards to an interface (a serial for example) which reencapsulate with a layer 2 frame. This serial has encapsulations set as PPP (or HDLC or whatever) therefore the layer 3 datagram is encapsulated in a PPP frame and forwarded out.

Therefore there is no such a thing like a layer 2 encapsulation in the Internet. The layer 2 encapsulation is associated to each and every link and changes whenever you cross a device.

For the second part of your question please refer to:

http://www.cisco.com/en/US/tech/tk648/tk365/technologies_tech_note09186a008009404a.shtml

As far as I am aware you cannot change the encapsulation of ethernet (using TCP/IP) but you can is you use IPX SPX (there are a few questions on it in the CCNA test).

HTH

Fabio

New Member

Re: tcp-ip encapsulation

first. thank you very much for your reply.

i didn't understand yet, what is the default encapuslation of tcp-ip in the

second layer of ethernet (10 base-t)

there are four type of ethenet encapsulation - 802.2,802.3,snap,ethernet II

when ip packet encapsulated by ethernet in the the second layer in which type it selcet. ?

802.3,802.2,snap,ethernet II ?

all i do is to configure tcp-ip in my computer,

when the ip packet is encapsuated in ethenet card, what will be the default

encapsulation ( it must be one of the four , or maybe there is a special encapsulation for tcp-ip packet )

?

thanks u very much for your concern

--bye--

Silver

Re: tcp-ip encapsulation

You seem to be c onfusing Novell NetWare IPX with Internet IP. Novell defined four modes of LAN encapsulation as part of configuring a Novell network:

1 - Ethernet - Uses Ethernet MAC encasulation with a protocol type of IPX. MTU is 1500 bytes.

2 - 802.3 - Uses IEEE802.3 encapsulation (same as Ethernet except instead of a type field, there is a length field). Since there is no mechanism to indicate that the frame is carrying IPX, this encapsulation is ONLY safe to use if ONLY IPX and no other protocols are ever present on the LAN. In other words, don't use it!

3 - 802.2 - Uses IEEE 802.3 MAC and IEEE 802.2 type 1 Logical Link Control (the only standard payload for an IEEE 802.3 frame). 802.2 identifies the payload of the frame as IPX by using a unique LSAP value assigned to IPX in the 802.2 header. This reduces the MTU from the 1500 bytes of the previous two down to 1497 bytes.

4 - SNAP - Is similar to 802.2 encapsulation except the LSAP used is "SNAP" which indicates that another five bytes of payload are used to encode the Ethernet type field as part of the 802.2 framing. You don't want to know why it takes 5 bytes to encode the two byte Ethertype, you only need to recognize that the MTU is now down to 1492 bytes. The good news is that any protocol which works over Ethernet can also work over IEEE 802 (as long as it can live with the reduced MTU).

IP encapsulations are defined for Ethernet (RFC894) and SNAP (RFC1042). Support for RFC894 is mandatory while SNAP is optional. OSI CLNS (may it rest in peace) can use Ethernet, 802.2 or SNAP. Novell's 802.3 is unique to Novell and is an abomination which should be avoided at all costs (yes, I am aware that hacks were developed to allow coexistance, but they require what would today be considered unacceptable compromises).

Good luck and have fun!

Vincent C Jones

www.networkingunlimited.com

441
Views
10
Helpful
3
Replies
CreatePlease login to create content