The ISP determines what your clock speed and duplex and that sort of thing is for your serial wan link. That then makes them the DCE end.
The link above is an example of a DTE,DCE cable
The above thread also goes into good detail that might help clear it up for you.
Also different routers can have 2 kinds of serial interfaces (as far as I know from my studies) theres a a T1 WIC which is the older bigger style with alot of pins, the WIC 2T I believe is the thinner one that sort of looks like a type of SCSI connection.
This link below will explain a WIC 2T interface card.
This one then explains the WIC 1T
As you can see, very different connections, but both are serial WAN interfaces. Its important to know the difference.
Another thread that might help with the CSU DSU Explaination -
Sorry about all the information - Hope it helps (please correct if I explain something wrong or overly complicated anything).
Mustafiz, whether using PT or real routers, when we connect routers like this, its called a serial back to back connection. Its only done in labs to simulate a WAN (its actually a kind of crossover cable). In the real world there would be some kind of intervening device, such as a CSU/DSU.
Its the CABLE itself that determines which end of the link is DCE or DTE.
So, in packet tracer do this: connect two routers using the serial cable that has the little clock icon on it. You don't need to add IP addresses, just issue the no shutdown command on each serial interface.
Now, on the side that has the little clock icon, run the show controllers command, like this:
Router#show controllers s0/0/0 (or whatever serial you have)
Hardware is PowerQUICC MPC860
DCE V.35, clock rate 2000000
Then run the same command on the other end of the link...what do you see?
Last, flip the cable and run the commands again.
The above link gives a really good amount of images that really explain a wan set up.
In short with that whole thread, you have a router(DTE) which connects to a CSU/DSU(DCE) via a v.35 synchronous serial cable. The router(DTE) takes the smaller male connector and the CSU/DSU(DCE) takes the bigger female connector, this connects the two together and the clock rate and other details are provided to the router(DTE) via the CSU/DSU(DCE). In the particular thread above, the CSU/DSU is then connected via a T1 service to an ISP, using an apparent RJ-48 connector which is a little different to an RJ-45.
Here is the image of it all together:
Hopefully this will shed some light!
Also: Thank you Keith Barker for this picture and his explination that I was able to refer to.
This is a back to back WAN connection using 2 routers on Packet Tracer as an example also. Kev is deffinitly right when using the cable can also determine who is DCE and who is DTE in a WAN link when its set up this way, I just chose R1 as mine, when making this sort of connection, you chose DCE Serial as the cable and click which router you want as the DCE and then connect to the other device as the DTE.
In a Back to Back Serial Connection, the Clock must be set at the DCE side, a source of clocking is always needed in Asynchronous Serial Link, because the routers are connected directly, its configured at the DCE side.
If the routers are not directly connected, ie: the Router is connected to the Telco CSU/DSU, no need for the Clock rate to be configured at the serial interface, as the Source of Clocking is provided by the CSU/DSU.
Thanks i got my answer for that...i got a new doubt
Let me define a situation wherin R1 and R2 are connected back to back serially and there is no other device other than these two..r1 and r2 are connected via v.35 cable,wher r1 is DCE end and r2 is DTE end(so r1 is clock source here)..
Now just tell me y do we need to name the link between them with a new Address?
Im sure theres more to the reasoning behind it that I know of, but my understanding is that if you dont have IP addressing over WAN, then the routing won't work between devices. You wont have a next hop IP when example RIPv2 attempts to find the best route between hosts. You can configure static routes without using a next hope by specifiying it to go out a specific interface, but I still believe it needs IP addressing to correctly work still. NAT wouldnt work either if you didnt have an IP address, example: if your network had a WAN Link directly to the ISP just for internet, if that WAN Didnt have an IP address you wouldnt be able to use NAT/PAT to access the internet from private IP addresses.
Again I'm sure theres a better reason than that but thats what comes to mind as what would happen if you didnt. A layer 3 device handles and requires layer 3 addressing.
Hope this helps,
[Please add to this if I've missed something or didnt explain it properly, as it would benefit me also]