Data Transfer Rates based on medium

[user125]

I had a small debate with a friend on transfer rates based on bandwidth and wanted some clarification.

Let say we have two isolated LANS.
one 100Mb Network and one 1Gb Network. (both copper)
if I transfer one file from point A to Point B using CIFS, would it take the same amount of time on a 100Mb network as it would a 1Gb network? (In theory) or, would it transfer faster in a Gigabit network? My argument was that regardless of CIFS' chatty nature, considering that in a gigabit network a signal can be received at a rate of 1 billion 1,0's per sec. i figured the request/response would occur much faster than it would in a 100Mb network. My colleague's argument is that because it is running on copper the time it would take for a packet to travel from point a to point b will always be the same and therefore regardless as to how fast your network is, in theory it will always arrive at the same time.

Is that true? Am i misunderstanding the concept of how fast a packet can travel? In school, my professor told me, that in copper wiring, the wiring is electrically charged on and off to signal a 0 or 1. So he goes on to say, if its a 10 Mb line, then a charge goes on and off 10 million times per second. SO, with that reasoning in mind, I figured if a gigabit network can transmit 1 billion 1,0's per sec, that that would mean one bit would be received at 1 billionith of a second because the actual electricity is going over a linear path.

 

[MasterRacker]

A bit "travels" from one end of a wire to the other at the speed of light in copper no matter what the cable rating is.

The difference is how long the bits are on or off. The cable rating is a function of how well it preserves the signal so shorter duration bits can be intrepreted correctly.

If you send a single bit down a 10Mb wire 100 ft long and it took 100ns to reach the far end, swapping out a 1000Mb will not change the travel time.

Jeff
[small][purple]It's never too early to begin preparing for [/purple]International Talk Like a Pirate Day
"The software I buy sucks, The software I write sucks. It's time to give up and have a beer..." - Me[/small]

 

[wires]

the time it would take for a packet to travel from point a to point b will always be the same and therefore regardless as to how fast your network is"

WRONG!

While the speed of an on-off transition is the same on any copper wire the speed at which PACKETS (of the same size) can be assembled, transported and received is quite another. A faster network protocol can send packets at a greater rate than a slower network. Otherwise what would be the point of a faster network?

MasterRacker is correct in the crux of the reason. It all has to do with how accurately a waveform can be preserved over a distance of a media (copper, fiber, RF...). Lower bandwidth networks tend to distort the transitions between high and low causing errors.

Is your friend argumentative or is he just challenged?

 

[MasterRacker]

Actually you could really mess with people here. It's all a question of accurate terminology. There is an actual measurable travel time from point A to point B and it's the same regardless of cable speed. However, the time to read a packet is shorter on a faster network, because the bits themselves have shorter duration.

Individual bits cannot be transmitted "faster", however data can be transmitted "faster" because the duration of each packet is shorter.

Now that we've cleared that up....

Jeff
[small][purple]It's never too early to begin preparing for [/purple]International Talk Like a Pirate Day
"The software I buy sucks, The software I write sucks. It's time to give up and have a beer..." - Me[/small]

 

[buitenhek]

Let's mention the cable:
The 1G cable was constructed from the current best-o-the-best materials and is capable of handling packets at 1G rates. The 100M cable can NOT handle data packets at 1G rates...

The amount of impurities in the copper cable "restrict" flow. Therefore, the magic happens in both the electronics and the cable manufacturing process. I know there is a lot of couth that goes with this but... both signals (1G and 100M) arrive at the same time, but only one can be interpreted correctly at 1G speeds.

Regards
Peter Buitenhek
ProfitDeveloper.com

"Never settle for a job well done...always look for cost cutting measures

 

[wires]

The amount of impurities in the copper cable "restrict" flow."

The "purity" of the copper has little to do with data rate.

The major issue is keeping the pulses free from distortion. The two main factors are electrical characteristics of the cable and crosstalk.

Electrical issues are mostly the effects of inductance and capacitance and how they effect pulse shape.

Crosstalk includes all the XT parameters you see on certification testers as well as alien crosstalk from other cables in the area.

An interesting read on 10Gb/sec and beyond on copper can be found here: ccswebapps.corning.com/web/library/litindex.nsf/$all/LAN-803-EN/$file/LAN-803-EN.pdf Of course they are selling fiber but the discussion of crosstalk and increased weight and power levels are interesting. With increasing power levels alien crosstalk becomes a larger issue.

 

[KTM72]

Each cable manufacturer publishes information, which is largely ignored, it is called the "NVP" or nominal velocity of propagation. This, in English, is the percentage of the speed of light at which a data bit will travel along the cable. This NVP is the part you need to look at regarding transmission speeds on different media. Amusingly enough Cat5e and Cat 6 share the same NVP on almost all mainstream cable brands. All "standard" types of media will have published NVP data if you look for it.

I was miserable, then someone told me "smile and be happy, things could be worse". So I smiled and was happy...and they were right, things were worse!

 

[wires]

Each cable manufacturer publishes information, which is largely ignored, it is called the "NVP" or nominal velocity of propagation."

It's not ignored if you use a Time Domain Reflectometer (TDR) since the NVP is how you translate time into distance. However to extrapolate data speeds in a ethernet network from NVP is foolish since the difference in NVP over 330' is negligible.