Packets or Frames

Ok I figured I'd make a post about this because I think sometimes "packets" or "frames" are used interchangeably when they really shouldn't be.

So I'm watching a video on STP and they are saying that the "packets" can continue on an endless loop unless STP is enabled.

Over and over, they use "packets" and not "frames".

I suppose if you want to get technical, the packets are encapsulated within frames but I really think one should use the correct terminology according to the layer in question.

To me, when you're talking about L2 switches and the topic of STP, it's all about frames isn't it? Aren't the frames the actual items that are being switched around endlessly?

When you talk about routers and L3, I don't hear anyone saying routers are forwarding frames, no they are pretty clear in saying packets.

So why are "packets" used when talking about L2 related topics?

Thoughts?

Find more posts tagged with

Free for TechExams community: Cybersecurity salary guide

Compare cert salaries and plan your next career move

Button

Comments

DevilWAH

switching and a switch are two different things, as is a router and routing..

switching and routing are technologies

on the other hand a SWITCH and a ROUTER are names of devices, and with in networking they use to have a defination, Switch was layer 2 and Router was layer 3.

In effect both switching and routing are the same thing, they both read a source address of some type or another and look up where it has to go in a table and forward it out a exit interface to the next device.

the term SWITCH was used to designate a device that did this at layer 2, and a Router did this at layer 3.

Now however there is so much over lap that they dont really exist as seperate devices any more.

I do think it can be confusing to people new to networking to hear the teams layer 3 switch or multi layer switch.

May be its just cause I rember when a ROuter was a ROUTER and a Switch was a SWITCH..

hehe

notgoing2fail

I have the book on IP CEF. I have to get it, there's a page where the author talks about not confusing the way a router switches and a switch switches.

That it is confusing but the processes are different.

I'll have to find it and see exactly what the author meant....

notgoing2fail

Here's the introduction to Cisco Express Forwarding book.

How does a router switch a packet? What is the difference between routing a packet, switching a frame, and packet switching?

notgoing2fail

OMG you guys are going to get a kick out of this. I just received my SWITCH book today and how timely is this! The author goes into a quick review on layer 2/3 switches. I pulled out just a couple sentences here but you can find it on page 8 and 9.

What a doosie!!!! You're going to fall off your chair!!!!!

Here's what the author says: My comments are in blue.

Layer 2 switches are capable of switching packets based only on MAC addresses.

I thought frames contained MAC address info? Not packets. Packets are Layer 3. Frames are Layer 2 and Segments are Layer 4.

So how can a packet contain MAC info in order to be switched?

The term Layer 2 switching implies that frames forwarded by the switch are not modified in any way.

Ok I can give the author this one. Usually packets are forwarded and frames are switched. But I'll throw a bone here.

Layer 2 only switches are not capable of routing frames based on IP address and are limited to forwarding frames only based on MAC addresss.

Of course not, frames don't contain any L3 address, so of course you can't route on IP. Again, he mentions frames are forwarded, not switched.....

Layer 2 switches might support features that read Layer 3 information of a frame for specific features.

Since when does a frame (a data-link layer) contain Layer 3 info?

Seriously, holy sh**, the author is ALL OVER THE PLACE.

Forget nitpicking, you guys have to admit, that's one confusing paragraph he wrote!!!!

networker050184

Makes sense to me. I think you are just nit picking way too much man.

fly351

My comments are in red

notgoing2fail wrote: »

Here's what the author says: My comments are in blue.

Layer 2 switches are capable of switching packets based only on MAC addresses.

I thought frames contained MAC address info? Not packets. Packets are Layer 3. Frames are Layer 2 and Segments are Layer 4.

So how can a packet contain MAC info in order to be switched?

Yes, the frame header contains the MAC address but the frame contains the Packet. What you quoted says "capable of switching packets", not that the packet contains the MAC address.

Layer 2 switches might support features that read Layer 3 information of a frame for specific features.

Since when does a frame (a data-link layer) contain Layer 3 info?

Doesn't QoS operate at Layer 2 and 3?

notgoing2fail

networker050184 wrote: »

Makes sense to me. I think you are just nit picking way too much man.

I know, it's my OCD kicking in.....

notgoing2fail

fly351 wrote: »

My comments are in red

Yes, the frame header contains the MAC address but the frame contains the Packet. What you quoted says "capable of switching packets", not that the packet contains the MAC address.

Doesn't QoS operate at Layer 2 and 3?

Ok for the first one, it could lead to different interpretations. My POV was that the author meant that the switch can make decisions on the "packet" based on the MAC address. Especially for Layer 2 switch, I don't know if a switch is capable of reading passed the frame and into the packet itself?

For the second one, QoS operates on both layers yes. CoS for layer 2 and DSCP/IntServ for layer 3.

So I'm ok with a switch reading layer 2 (CoS) frame info. But to say that a switch reads layer 3 frame info is nonsense. It's just sloppy authoring. There's no such thing as a layer 3 frame!!!

networker050184

So if there is no Layer 3 info in the frame how is it routed when the L2 header is stripped off?

notgoing2fail

networker050184 wrote: »

So if there is no Layer 3 info in the frame how is it routed when the L2 header is stripped off?

No that's not what I meant, what you just said is what I was trying to convey.

That layer 3 strips off frame info, therefore you are left with packet or datagram.

What I'm trying to say is that a packet can be contained within a frame on layer 2, but frames don't exist on layer 3.

networker050184

What if its a frame being tunneld over IP or MPLS? I think you need to just let it go man

notgoing2fail

networker050184 wrote: »

What if its a frame being tunneld over IP or MPLS? I think you need to just let it go man

Then you have got yourself a good point. That is interesting indeed....

I have not thought about tunneling...

DevilWAH

and how about a packet capture/anaylizer program like wire shark??

should this not be a frame capture application

HardDisk

Let me get this straight.

A layer 4 Segment is "routed" once it is encapsulated within an IP Packet aka DataGram.

An IP Packet is "switched" once it is encapsulated within an Ethernet Frame.

So can I then say that a TCP Segment is both "routed" and "switched"? (rhetorical question).

Enjoyed this post with my morning coffee. Thanks

DevilWAH

HardDisk wrote: »

Let me get this straight.

A layer 4 Segment is "routed" once it is encapsulated within an IP Packet aka DataGram.

An IP Packet is "switched" once it is encapsulated within an Ethernet Frame.

So can I then say that a TCP Segment is both "routed" and "switched"? (rhetorical question).

Enjoyed this post with my morning coffee. Thanks

Well only routed and switched if it passes across a router. if its only passes across a layer 2 network it is never routed

I wonder how may new network people we have discouraged so far from getting in to this field

fly351

DevilWAH wrote: »

I wonder how may new network people we have discouraged so far from getting in to this field

hahaha I am sure this thread is got some new people going

HardDisk

If you made it past the fact that the ISO created the OSI model and that ironically somehow ISO stands for "International Organization for Standardization" then you've got what it takes.

notgoing2fail

DevilWAH wrote: »

Well only routed and switched if it passes across a router. if its only passes across a layer 2 network it is never routed

I wonder how may new network people we have discouraged so far from getting in to this field

Right, you can say if the destination IP address is local to the subnet it's just switched...

Let's hope no one turns away from this field because of this thread. I really think it's a thread worth discussing.

Why shy away from it? Let's get to the bottom of things, is it nitpicking? Yes.....

But it's also a little fun too ...... I can't be the only one that's a little curious how the terms are thrown around loosely....

fly351

They might be loosely thrown around, but at the same time... if someone is working on their CCNP I hope they can understand the difference

I can see where someone would get confused at the CCENT/CCNA level, but if you can grasp the different encapsulations then it shouldn't be to confusing.

notgoing2fail

fly351 wrote: »

They might be loosely thrown around, but at the same time... if someone is working on their CCNP I hope they can understand the difference I can see where someone would get confused at the CCENT/CCNA level, but if you can grasp the different encapsulations then it shouldn't be to confusing.

Absolutely. I think that is actually why CCIE's/video/book authors loosely throw the terms around because THEY know what they are really talking about, and if you're a CCNP wanna-be reading their books, you should too.

The up and coming engineers may get slightly confused.

burbankmarc

Routing involves a RIB lookup where switching is cache based. There seems to be a fair amount of over thinking going on here.

DevilWAH

notgoing2fail wrote: »

Absolutely. I think that is actually why CCIE's/video/book authors loosely throw the terms around because THEY know what they are really talking about, and if you're a CCNP wanna-be reading their books, you should too.

I think the real main reason, is that many of the terms such as a Switch and a Router are so out of date now that it becomes inpossible to use them.

Things like the OSI model as well are really just guides and there are so many exception to the rules.

I mean surely a router should also be called a fire wall as it can filter packets based on policies, and what about fire walls that can do routing?

What I mean is at what point does a router become termed as a firewall or a fire wall termed as a router?

notgoing2fail

burbankmarc wrote: »

Routing involves a RIB lookup where switching is cache based. There seems to be a fair amount of over thinking going on here.

Just trying to have a healthy debate....

DevilWAH wrote: »

I think the real main reason, is that many of the terms such as a Switch and a Router are so out of date now that it becomes inpossible to use them.

Things like the OSI model as well are really just guides and there are so many exception to the rules.

I mean surely a router should also be called a fire wall as it can filter packets based on policies, and what about fire walls that can do routing?

What I mean is at what point does a router become termed as a firewall or a fire wall termed as a router?

Yeah this is true. The ISR's have really changed what is a router, what is a firewall and what is a switch....oh, and what is an IPS!!

I do know that it also comes down to how the device deals with the packets/frames.

For example an old school router will forward packets done at the software level so this is true forwarding and more CPU intensive.

But since "ip route cache/IP CEF", now you've got FIB tables that are clone copies of routing tables. So the first packet goes to the route processor, and all ensuing packets are now switched at the hardware layer.

I suppose this thread should really go into the CCNP section but I think it's healthy to talk about for new CCNA's.

Is it overthinking/nitpicking with some OCD mixed in? Sure!!

But why not discuss and see what other people think? That's why I created this thread....I would love to get CCIE point of views on this....

alan2308

notgoing2fail wrote: »

Is it overthinking/nitpicking with some OCD mixed in? Sure!!

That's what we do here.

notgoing2fail

alan2308 wrote: »

That's what we do here.

There was a VLAN lab that I did recently, haven't finished it yet, but boy, it totally wakes you up as to your knowledge of VLAN's....

Just when you thought you knew....

alan2308

notgoing2fail wrote: »

There was a VLAN lab that I did recently, haven't finished it yet, but boy, it totally wakes you up as to your knowledge of VLAN's....

Just when you thought you knew....

One of the free labs Narbik Kocharians gave away was a lab on RIP that is supposed to take someone with CCIE level knowledge 8 hours to complete. Just looking through it is quite a humbling experience.

mgeorge

Notgoing2fail; you're definitely on the right track with understanding switching/forwarding/cef and you're right although many engineers and authors loosely throw around the terms. Either because of their lack of intelligence or lack of motivation to explain the stuff in detail lol.

When thinking of stuff like this you really have to reference the OSI model (yeah that annoying ass 7 layer piece of paper that you'd thought you'd never use in real life) haha...

First off we all know layer 3 uses packets as the pdu, layer 2 is frames and layer 1 is the bits... (keep in mind 8 brownie bits equal a brownie byte) haha. I could totally go for a double chocolate fudge brownie right now, anywho...

You have to keep in mind that the OSI model is like the harmony in the matrix, one layer dependent upon another layer and shall one fail the entire matrix will fail haha.

Any PC sending traffic to another PC on the same network is going to be in some form or fashion switched (unless you use a hub). If its the same subnet then the PC will ARP for the mac address of that destination IP and place it in the DST field of the frame and send it on the wire to the switch. However if the PC knows that the destination IP address is not in its local subnet it sends the packet out the wire with the mac address of its default gateway.

This is all common knowledge. When the frame is received by the router, because in order for a device to process incoming traffic it has to go back UP the OSI model, not just start at layer 3 like some people magically believe. It then examines the packet contained in the frame (kinda like a picture in a picture in a picture frame) and then checks out layer 4 and 5 for other ip services related stuff but ultimately determined rather or not that particular packet needs to be CEF switched or process switched. If it is CEF then the router references the CEF table (aka: forwarding information base) and then rp recalculates the CRC and then forwards the packet (yes forwards it as its switched via SW or HW ASIC's in the router) out a particular interface with the matching route prefix. If it is process switched for whatever reason such as ACL logging or policy based routing, NAT, etc... then the route processor process any policies configured for the traffic then look up the routing table and forward the traffic accordingly out a particular interface tied to the route.

Keep in mind Cisco is becoming good at offloading process based functions via ASIC's, there are a lot of Cisco devices that do NAT, PBR and ACL processing using hardware. In this case the same functions still occur, its just done by a different processor (ASIC's) so to speak. In some cases where all HW based processing is used, the RP may not be used at all (which is Cisco's goal), traffic comes into a layer 3 switch, goes up and does its little dance with the ACL/PBR/NAT/CEF ACISs and gets shoved off the stage at the local county fair ho down dance.

(meaning forwarded out the correct interface)

A router performs functions of a layer 2 device even though its main purposes is layer 3/4. How else does traffic get from layer 3 back down to layer 2 then forwarded over a layer 1 medium? It's not magic and you cant blame it on the matrix either haha.

Layer 2 switches switch the frames based in src and dst mac addresses. Layer 3 switches basically combine layer 2 and layer 3 functions in a single box obviously, in this case any PC sending traffic to a PC on a different subnet forwards the frame out the PC's NIC through layer 1 with the dst mac address of the switches RP. Once the RP does its thing, it hands it back down to the SP then gets switched out the specific interface where the dst mac address is located at

As far as the terms go, if you understand the operation of the OSI model and the technology that Cisco uses to accomplish the requirements of the OSI model put forth by the International Standards Organization then you should not have a problem with understanding this stuff ^_^

Spanning tree in and of its self is used to prevent multiple active physical/logical links between two devices on a common layer 2 network segment by blocking the forwarding path of one link to break the loop.

Switching - when a frame is switched from one interface to another on a switch based on source and destination mac addresses.

Forwarding - This term is loosely used to describe an action taken by a device to forward the traffic it receives along the transit path from the source to the destination regardless of how the traffic is processed.

CEF Switched - This differs from Layer 2 switching. The router keeps a mirrored copy of the RIB (routing information base; aka: routing table) and calls it the FIB (forwarding information base; aka: cef table) and uses the CEF table as a layer 3 switching table so when traffic enters the router, if it is not process switched it is cef switched. The router will look up the destination ip in the cef table and forward it out the attached interface.

As an instructor it amazes me to see how many people just memorize the OSI model and what happens at each layer and goes on with their happy network engineering life without understanding how each layer inter-operates with each other. The comprehension of the OSI models lower layers if critical to a fundamental understanding of basic network operations.

Sorry for the long reply, just felt the need to rant...

mikej412

mgeorge wrote: »

double chocolate fudge brownie

This is all I saw in your post.....

Just kidding -- nice post

mgeorge

Haha!!! I knew that was coming from someone, I was waiting for that reply...

The thought alone is mouth watering isn't it?

Although I forgot to add the sprinkles... Gotta have the sprinkles ^_^

There is a great book in the Google Books library that discusses the different types of switching; check it out if you want an in-depth analysis of all the crazy operations of switching types.

CCIE practical studies, Volume 2 - Page 308

burbankmarc

mgeorge wrote: »

There is a great book in the Google Books library that discusses the different types of switching; check it out if you want an in-depth analysis of all the crazy operations of switching types.

CCIE practical studies, Volume 2 - Page 308

This link has info on the different switching types too, except for flow based switching since it isn't used anymore:

Cisco IOS Switching Services Configuration Guide, Release 12.2 - Cisco IOS Switching Paths Overview [Cisco IOS Software Releases 12.2 Mainline] - Cisco Systems