packet vs frame

Packet:

Frame:

Basically, a frame is a collection of one or more packets.

Hmmm. I disagree - certainly in the conext of alt.internet.wireless.

Different people use these terms casually to mean different things and I would be surprised if different Standard Bodies had not specified - carefully - multiple different meanings for each of them.

I would not dare to attempt a definition since I would be bound to be wrong.

The one thing that I am reasonably confortable with is that I have never seen a frame described as Jeff has described it here. : -)

So, to address the question -

Frame means very different things in the T1, T3 and perhaps DSL world than it does in the Ethernet world.

What I would do is to always specify the meaning of packet and frame whenever used to ensure that there was no ambiguity if that was appropriate.

e.g. the packet length was 234 bytes (including data, tcp and ip headers but excluding Ethernet encapsulation).

I think that in the Ethernet World, a Packet is everything except the Ethernet headers/trailer and frame includes all of the Ethernet bits too. Those being at least Destination address, Source Address, Length/type, 802.2 and SNAP if present and CRC. Frame may or may not include the Preamble.

Oh dear definitaion attempted and the only sure thing is that it is definately wrong:-)

No real difference until you get pretty close to the wire-level electronics, where the 802.3 specs say (in specmanslike gory detail) that: a data frame is contained within a packet for transmission over 802 networks. I.e., a data frame becomes a packet by tacking on some control bits fore and aft.

There are, of course, some packets that do not contain data; part of the overhead tax.

Alt.internet.wireless has context? I hadn't noticed.

Yep. I tried to supply a simplified explanation and apparently failed. You opened this can of worms, so you get the long version.

Bingo. Telco buzzwords are very different from computah buzzwords and for good reason. Back in the daze of Ma Bell, AT&T was very careful not to give the impression that they were a monopoly. A monopoly was a company that manufactured and sold at every level of the industry, from raw materials to final delivery to the consumer. It was therefore necessary for one part of the puzzle to be left untouched by the telcos, and that was computahs. Ma Bell was very careful not to give the impression that they were in the computer business. This was possible at the time because computers were in their infancy and not widely used.

However, Ma Bell just couldn't ignore computers or resist the temptation to use and sell them. The best she could do was to obscure and camouflage the terms and buzzwords that surround computers. For example, a computer was called a "switch". Many terms as possible became acronyms, which further obscured the underlying technology. Early versions of Newton's Telecom Dictionary underscored the equivalent terms. At the time, I worked for Ma Bell, and was going to skool learning on an IBM 1620. I went through a vocabulary and personality change as I switched from telco to computah jargon. At the CO, we were warned the using computer terms was "dangerous".

When working for Ma Bell, the TDM systems of the day sent frames. In the computers of the day, there were no packets. However, to distinguish between what was digestible by the later computers from that which was handled by the telco, the delivered payload were called packets. In other words, when the data was on the phone lines, it was in frames. When coming out of the CSU/DSU, it magically became packets.

In 1984, it became a non-issue as the government did the stockholders a huge favor and broke up Ma Bell into the Baby Bell's. Now, there was no longer an issue of being called a monopoly and the Baby Bell's could easily build and sell computers. Slowly, the terms merged, were standardized, conglomerated, and were of course, misused by both the computer and the telco crowd.

So, when you discuss data communications with a computah person, it's a packet. When you discuss data transmission with a telco person, it's a frame.

That's not quite right. Actually what you're describing is a vertically integrated manufacturer/distributor.

A monopoly refers to having the entire market to yourself in an area-- say like the gas company, cable company, electric company. It doesn't matter where you get the stuff or who makes it. For example, the local cable company buys content from the networks and others, buys the wires they string to your house-- any maybe even sub out the installation--- but the cable company is a monopoly in your area because you can't buy that signal/content from any one else (disregard satellite for the purpose of this example).

hmmm - OSI model - thinking the above layers add "protocol" bytes and assemble/dis-assemble "packets" X.25 and PAD's - is there a corresponding FAD ? (not FRAD) while the lower layer merely adds timing to the "frame", hmmm - but what about 56k vs 64k and CRC and B8ZS, and .... oh heck -

Right. I stand corrected. That's what I meant to say. AT&T spent considerable effort to make sure that it did NOT manufacture, sell, distribute, and dominate every area of the telephone business. In about 1958(?), there was some court ruling that prevented AT&T from going into the computer business (mostly thanks to pressure from IBM). Here are the details:

Ma Bell turned this ruling to its advantage by convincing subsequent anti-trust judges that the lack of involvement in computers, constituted evidence that they did not have a vertical monopoly. It worked, but required that nobody notice that their billing machines were nothing more than camouflaged DEC PDP-11 minis, and that their

5ESS switches were really run by Unix based general purpose computers. So, they simply changed all the name to confuse the clueless.

Right. As I understand it, the courts and law manufacturers have nothing against creating a monopoly. What is illegal is using that monopoly to gain some advantage at the expense of the competition. Kinda like what AT&T is currently doing to their former DSL "partners" by wholesale DSL service to their partners for more than they're selling the same service directly to the customers.

Argh. I'm late....gone.

don escribió:

IMHO, I use term "frame" when I talk in Link Layer context, and I use term "packet" when I talk in Network Layer context. Of course, I am talking about ISO/OSI model.

So, for me, a "packet" is a Layer 3 PDU (protocol data unit) and "frame" is a Layer 2 PDU, so packets are encapsulated into frames.

Of course, ISO/OSI model, as its name says, is just a model, and you can be agree or disagree with it, but I think it's a good starting point to understand network mechanics.

Have a nice day. Angel Catala.

Ethernet sends frames.

IP sends packets.

TCP sends segments.

A more generic word for things like frames, packets and segments is a "datagram".

If you're communicating with another computer using TCP, then you're sending frames which contain packets which contain segments.

I think recently though people have started using "packet" to refer to a frame.

I think I know the computer definition, is a packet is Layer 3 of the OSI, and a frame is layer 2 of the OSI. Perhaps the terminology is lacking, it just refers to different portions of this one unnamed thing!!! And technically, I am not even sure that Frame is said to include the Packet. I think it just refers to the frame fields.. And they are just drawn with this packet, L2 SDU / L3 PDU thing in it.

I am not sure about the telecommunications definition.. I had heard that Packet , as in "packet switching", refers to the thing on the wire.. Lower level than frame. Just the bits, or signals even, whatever they are. Didn`t know it can refer to many packets though.

I read on usenet - I think from rich seifert - that in OSI terminology, the the PDU at the physical layer is called a symbol!!!!!!! Of cousre nobody uses that word though..

Once on the wire, indeed, people use telecommunications terminology.

But prior to that / at the computer level, The OSI terminology is used.. But the term Packet is still used sometimes to refer to the whole thing. At the computer level. Probably because there does not seem to be an OSI word for the whole thing (though I have not read OSI!)

Maybe, when you are referring to Bits, then you are at computer level.. When on the wire, you are talking signals, then it`s OSI.. I think that`s the way the OSI looks at it.. Maybe telecommunications look at it differently, seeing themselves as dealing with bits too!

away from questions I have....

This may help.. I don`t like this book much but it has its good aspects.. it draws from many sources(of which the author gives no references!). it can resolve differences between materials..

" Packet: This term is considered by many to most correctly refer to a message sent by protocols operating at the network layer of the OSI Reference Model. So, you will commonly see people refer to "IP packets". However, this term is commonly also used to refer generically to any type of message, as I mentioned at the start of this topic. "

The words packet and frame, if used together, then certainly the word packet is not in its most general form. It is specifically to the part of the "thing", the packet part, it has the IP fields. and the word Frame, refers to the part that has the frame fields.

the other way they look at it is not as fields, but as containers. So fields with the SDU.. So the whole thing from bits, starts at Layer 2(data link layer), consists of the frame fields, followed by the Layer 2 SDU, which is the Layer 3 PDU, which is the packet. The Packet is the Layer 3 PDU(network layer), which consists of Layer 3 fields, followed by the Layer 3 SDU, which is the Layer 4 PDU (transport layer), which is the TCP segment with the....

The thing is.. OSI defines 7 layers in its reference model.. But we use that terminology on the TCP/IP architecture.. and TCP/IP only has

5 layers. So skip session layer and presentation layer (layers 5,6). And we get Layer 7.

One could read about this.. cisco ccna books touch on it.. TCP/IP books. I have a book called "using TCP/IP" by que.. A great one is Networking first step by wendell odom. another one was windows 2000 for dummies, which had some good stuff on networking before it got to windows 2000... I actually cannot remember which books I used to learn about the OSI.. May have been largely the internet. some pages of that TCP/IP guide maybe.. I recall seeing a page of it with a picture.. this may have been it, it explains the terminology a bit.. And that helps for understanding other sites.

For learning, I rather like the 1988 book by Douglas comer, entitled "Internetworking with TCP/IP".

comer , or the other tomb, stevens,

are quite hardcore multi-volume series.. I haven`t read them. But to know what a packet is and what a frame is, they are probably overkill.