Bandwidth/Speed

To the experts,

I have a bandwidth question relating to speed. Given the following scenario I'm uncertain as to why this works out the way it does. Any information provided would be a great help.

If host X pings Host A using a dial-up connection (ppp), why is there a longer delay then if Host X pings Host A using a high speed lan connection (100mb)? The ICMP packet itself is small enough to fit on both dialup and highspeed copper wires without fragmenting and the electrons all travel over the "copper" at the same rate so why would there be any difference in delay speeds vs. the two connection technologies?

I ask because I see high ping rates from dialup users vs. highspeed users in relation to online gaming.

Thanks, Carl

Reply to
Carl Lovejoy
Loading thread data ...

dial up modems use the standard telephone lines. It's slower. Not really to do with electrons movign slower. i'll explain in a moment.... (by the way, I only learnt this a few days ago after much googling and speaking to electronics people).

You're right that fragmentation isn't the reason.

I'd have to meditate over it for a long time to see where speed of electrons comes in, but it is not any part of the reason.

Look at the signal as a wave. up and down and up and down with time. THe more ups and downs it does per second, the higher the frequency. (up and down is voltage, y axis. x axis is time) Dial up uses the max allowed frequency for telephone calls, that is,

4KHz. I guess coppor would of course allow higher frequencies, but for some reason, with high frequencies , there is noise and it doesn't come out so good. By the way, If a girl screams down a telephone line above 4KHz it 'cuts out'-apparently.

Digital breaks the old rule, uses frequencies above 4KHz. (it only uses this old technology till it gets to the telephone company, so noise distortion isn't such an issue - don't quote me) Don't quote me, but It's not coppor wire all the way. It is analogue a short distance, to the telephone company, then becomes a digital signal to the other guy's telephone company, then it's analogue again. the digital signal isn't sent over coppor, I don't think. It's sent over a medium that doesn't have noise issues distorting the signal when sent at high frequencies.

But that's the reason. why dial up is slower than broadband. The wave is sent at a lower frequency. BTW. An analogue cycle of a wave represents many bits. A digital only one. With fast DSL technologies, I don't think they use coppor wire, they get less noise, so can send at extremely high frequencies. I don't know why they don't use analogue signal all the way, since each cycle represents more than one bit. I guess because digital is quicker to read.

There may be many mistakes. I cut my explanation because had I written more completely on topics I know so little about, I'd have made many mistakes than perhaps I already have, and that'd put a burden on knowledgeable contributors.

Unfortunately, I don't have a background in electronics

It could be that a cycle of an analogue wave doesn't actually represent many bits. But there are 'many waves' sent down the wire, this is done by somehow combining them into one wave, and then extracting all the waves at the other end. (frquency multiplexing) It might be only broadband analogue waves that do this. Though I think dial up isn't baseband , but I don't think it's broadband. I don't know. This is obviously beyond Cisco (at least beyond CCNA!) , but is very interesting.

Main point, is it's not about the speed at which the electroncs move. It's the speed of the wave. THe speed at which the signal oscillates. And Noise is what prevents sending at too high a frequency. by the way. The Up and down of the wave - that is the wave's amplitude, that is voltage, it is the y axis. The x axis is time. (I know I mentioned that snippet already, but it's key).

Nothing to do with protocols like PPP or ICMP or Fragmentation of IP packets.

Reply to
q_q_anonymous

Carl:

The answer to your question is "serialization delay."

It takes longer to clock a packet out of the transmit buffer onto the wire on dial-up line (56k) than it does on a LAN segment (min 10mbps).

And if you think about it, the receiving end has to clock the packet into the receive buffer as well.

Naturally serialization delay will be shorter for higher speed connections.

Also, on slower links the packet takes up more "space" (i.e., be longer) on the wire, since the modulation intervals are longer on a slower speed link.

The same sized packet would be "smaller" (i.e., shorter) on the wire on a higher speed link because the modulation interval will be shorter.

Derick

Reply to
Derick Winkworth

Awesome, thanks for all the responses. I understand. I knew there was a reason I just couldn't figure it out.

Reply to
Carl Lovejoy

It is called serialization delay, it is the time it takes to put bits on the wire...

For example, it takes 214ms to put a 1500-byte packet onto a 56kbps link

12000 (8*1500) divided by 56000 = 214ms

It takes .00012ms to put that same packet on a 100mb link

12000 divided by 100,000,000 = .00012ms

Jonathan

Reply to
Jonathan

just to add. a higher freq , more cycles per second. cycles represent bits. (1 bit if digital. many if analogue) So, the higher the freq, the more bits on the wire per second, i.e. more Transmission speed.

Transmission speed is bits per second. e.g. 56k, The higher the frequency, the smaller what some have called 'serialization delay' The 802.3 spec calls the transmission speed the bit rate. And the 'serialization delay' it calls 'bit time'. Bit time is the inverse of bit delay. i.e. 1 divided by. serialisation delay is the inverse of transmission speed.

56Kbps transmission, means each bit takes 1/56k

I hope that links together what I said to what others have said.

So, what I said about the frequencies, corroborates what has been said about seriailization delay. One is the inverse of the other. One is bits per second. The other is seconds per bit.

I'm gald we got uniform responses!

Reply to
q_q_anonymous

Cabling-Design.com Forums website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.