Clock rate question.

pizzafart

This is something I suppose I should know by now, but, I don't understand how clock rates correlate to the bandwith available on a link.

For example, I'm consoled into a 2501 router on which there is a serial interface with a DB60 socket. The default bandwidth value on the interface is 1544Kbps. So, I assume that this is a DS1/T1 rated line. I could probably check the hardware specs on this model to be certain, but, I figure this is a safe assumption. In interface mode, there's no clock rate value that equals 1544000. How come? The closest values are 1300000 and 2000000. What gives? I've always used 2000000 in a lab environment though I wonder if this will be adequate in a real WAN situation where the DB60 cable converts to a different sort of connecter at the DCE.


This is for BCRAN studies... any help is appreciated!
Allan

BubbaJ

The clock rate is the actual line speed. The bandwidth command actually hs no effect on the controller or interface. It is a reference applied to the interface that other things can use. For example, EIGRP, by default, is allowed to use 50% of the bandwidth, and you can change the percent in EIGRP. EIGRP bases this on the bandwidth command even though that may not have a real relationship to the actual bandwidth that the interface is using.

pizzafart

I understand that the bandwidth value itself is just a software marker. But, isn't real T1 bandwidth usually rated at 1544000bps? 24 DS0's = 64kbps * 24 = 1544kbps = 1544000bps.

So, does that mean that this router is unable to clock at a regular T1 speed? That seems unusual considering that a T1 line would probably have been a popular choice for this sort of interface. I do appreciate the feedback but I'm still a bit confused.

BubbaJ

pizzafart wrote:

So, does that mean that this router is unable to clock at a regular T1 speed? I do appreciate the feedback but I'm still a bit confused.

Well, you are still missing a piece to connect to a T1 line: the CSU/DSU. The CSU/DSU will provide the clocking to the router - it will be the DCE while the router is the DTE. The line rate is set on the DCE, or comes from the line to the DCE, which will tell the DTE (router) what the line rate is.

The clock rate that you are referring to would be used in the case of the router being the DCE such as back-to-back router connections.

pizzafart

Aha, that may clear things up for me.

So, the values I'm seeing are probably geared towards the overall capability of that interface as opposed to being specifically tailored for an incoming T1 line. It is also notable that the 2500 isn't going to be a DCE for a T1 in the first place.


thanks for your help!

EdTheLad

BubbaJ wrote:

The clock rate is the actual line speed. The bandwidth command actually hs no effect on the controller or interface.

It's funny i've been thinking about the clock rate recently.
For me the clock rate controls the serialization speed rather than the line speed.Its funny, i always envisioned the actual bits propagating the line at the clock rate, while actually the electrical signals travel at roughly 2.1 x 10^8 m/s (a little less than the speed of light).So the clock rate is actually how fast the interface can process the data rather than the actual speed of the line.It doesnt really mater which ever way you view it generally but its better to know the true nature of the transmission in my opinion.
Anyway line speed is a quick easy analogy.
Has anyone else thaught about this before or am i losing the plot?

TheShadow

I look at it a little finer. While it is true that the signal is traveling at almost the speed of light it is the definition of signal that needs fine tuning. It is actually the electrons, the basis of a signal, that travels at ~ the speed of light. Clock speed is how long do the electrons travel at that speed. We normally just refer to that as pulse width/duty cycle which is normally expressed as a percentage for digital pulses. Clock speed then represents how many pulses of electrons per unit time are allowed or with fiber optics, how many light pulses per unit time are allowed. This then is simply how often do we measure.

Danman32

True, electrons flow nearly at the speed of light, but when you are talking about signals, how those electrons flow (mostly quantity) is what matters. For AC signals, the flow actually reverses direction! Because of the physical characteristics of the wire, if the signal state changes too fast, it gets attenuated and reshaped so that the signal no longer is recognizable as the original. So the clock rate (and signal change rate) regulates the speed of signal change so that the signal doesn't deteriorate.

Most cars can do about 100 miles an hour or better, but at that speed they are hard to manage, especially on sharp turns. That's why the speed limit on roads is much lower, so the car (the signal) and other cars (other signals) don't get damaged in transit.

EdTheLad

Danman32 wrote:

True, electrons flow nearly at the speed of light, but when you are talking about signals, how those electrons flow (mostly quantity) is what matters. For AC signals, the flow actually reverses direction! Because of the physical characteristics of the wire, if the signal state changes too fast, it gets attenuated and reshaped so that the signal no longer is recognizable as the original. So the clock rate (and signal change rate) regulates the speed of signal change so that the signal doesn't deteriorate.

Most cars can do about 100 miles an hour or better, but at that speed they are hard to manage, especially on sharp turns. That's why the speed limit on roads is much lower, so the car (the signal) and other cars (other signals) don't get damaged in transit.

Ok, TheShadow, i cant really understand your explanation.Danman32 i understand what you mean about the clockrate being used to determine which signal value is correct due to physical variations in the signal, it all comes down to rise edges of the clock and so on.So the speed limitation all comes down to how long it takes for the clock to transition from + to - , if the clock transition is too small physical errors on the line could cause a clock transition i guess.So as i see it depending on how you set the clock rate varies how long the clock stays in the same state without changing.But even going down into the finer analysis the clock rate must control the data being sent to ensure both are synched correctly,so on a serial interface its more the hardware limitation of the router interface which fixes the flow to max at T1 rather than the electrical characteristics of the line.It would be interesting to find out what the potential max value for transmission of data on copper really is, to do this you would need to know the minimum width required for the clock pulse.Maybe i should start reading the electrical specs hmm maybe i have too much to read already... Cheers for the feedback guys.

BubbaJ

ed_the_lad wrote:

It would be interesting to find out what the potential max value for transmission of data on copper really is...

In theory, it is one bit per electron. You could use spin to determine on or off. In practice, it increases all the time.

Danman32

Not quite, Ed. The clock determines the rate that the data will be accepted. The physical characteristics of the electronics and cable media determine the max data transition they can handle.

Take CAT3. You can't pump 100Mb through it, you'll get bit errors. I know, I have inadvertently tried. If you scoped the signal coming through the CAT3 at 100Mb, you'd see the distortion of the signal.

In case you are wondering how I inadvertently placed 100Mb on CAT3, my dad wanted to wire his and my mom's computers together. We tried HomeLAN which uses the phone lines, but that failed miserably. My dad had some network cable he obtained when he worked for a now defunct airline. I had some older network cards I gave him. That worked find for Win98, but when we upgraded to XP, we could not get drivers for the NICs. We bought new NICs that were 10/100, but forgot about the cable type now going through the walls and attic. Got transmission failures galore that I originally attributed to caching and name resolution problems. Then it hit me. WE HAVE CAT3! I forced the link speed down to 10Mb and it worked perfectly. My dad ran down to home depot, got some CAT5, rewired the LAN, and was able to communicate at 100Mb after that.

TheShadow

Danman32 wrote:

True, electrons flow nearly at the speed of light, but when you are talking about signals, how those electrons flow (mostly quantity) is what matters. For AC signals, the flow actually reverses direction! Because of the physical characteristics of the wire, if the signal state changes too fast, it gets attenuated and reshaped so that the signal no longer is recognizable as the original. So the clock rate (and signal change rate) regulates the speed of signal change so that the signal doesn't deteriorate.

Most cars can do about 100 miles an hour or better, but at that speed they are hard to manage, especially on sharp turns. That's why the speed limit on roads is much lower, so the car (the signal) and other cars (other signals) don't get damaged in transit.

Sorry Danman32, of course we are talking most quantity I did not want to go too deep and start talking about coulombs and what is a volt that would just confuse everyone. Your analogy is much better. If I am teaching someone I normally describe how the definitions Mark and Space used in telecommunications came about first.

TheShadow

You can buy cat5 at home depot? I would have never thought of that, must remember for the future.

pizzafart

Yikes, this is going a bit over my head.
Just to be sure that I have a fundamental understanding, could someone verify that the following statement is true:

If a Cisco interface is clocked at 64000, assuming that the CPU and physical media is in spec; this means that 64000 bits per second are sent/received at this interface. So, when I connect my 2500 routers back to back using a DCE/DTE cable and I set one of the interfaces to clock at 2000000; this means that I can send at most 2mbps of data through that line.

This discussion is interesting but I just want to be sure that I'm standing on a practical foundation.
Don't feel obliged to answer, though, I wouldn't mind some help.

EdTheLad

Yes thats correct Pizza, the data transferred matches the speed as the clock.

BubbaJ

pizzafart wrote:

If a Cisco interface is clocked at 64000, assuming that the CPU and physical media is in spec; this means that 64000 bits per second are sent/received at this interface. So, when I connect my 2500 routers back to back using a DCE/DTE cable and I set one of the interfaces to clock at 2000000; this means that I can send at most 2mbps of data through that line.

Yes, but it has to be the DCE end. DTE detects the speed on the line.

pizzafart

Cool. I'm glad to see this thread generated so much discussion!
It makes sense to me now. Actually, I took a closer look at some of the clocking/electrons posts above and they too are making more sense to me.

thanks folks!


btw I'm interviewing for a WAN position at Limited Brands. It just so happens that I'm currently studying for the BCRAN (I kind of wished I picked it earlier). Well, I came clean and told them that I had "limited" WAN experience. Though, it seems that they're still somewhat interested... so, I'm cramming hard partly in preparation for the bcran, partly in hopes that I can show my determination in being a solid wan admin. *fingers crossed*

This could be my first solo networking job (currently I intern with the network services department at a hospital).

Anyway, I just figured I'd tell someone. The folks I intern with are mostly approaching retirement and aren't very into expanding their skill sets. It's sort of silly but I often find that people who post on this board are more indentifiable as my "peers" than the folks I work with. Kind of sad I guess? lol, nah