Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Execute/Dumbo.exe]

So, my brother has Fiberoptic cable in his house, and he's said that his download speed can get up to 2-3 megabits per second (We live in Australia, alright?) but then asked me why its download speed is still weirdly slow, because if it uses light to transmit data, then the data should be transferred at light speed, right? And now I'm confused about that, too, why is it so slow if it uses something so fast to transmit data?

[TheBiggerFish]

Because hard drives use mechanical parts and have limited capacity to actually read/write the data they're getting/sending.

And there's a whole bunch of people using that data line, too, and it has a finite capacity.

Ask in the Generic Computer Advice Thread in Life Advice, maybe.

[TheBiggerFish]

Fiber optic cables aren't hollow is why. The light is still traveling through a transparent medium inside, and as you've observed at least once in your life, light travels slower in a non-vacuum medium (drinking straw bends in water).

Technically true.  That's not the limiting factor on his download speed though.

[TheBiggerFish]

Well, it was to the second half of his question. Or at least, the "light as a medium" part.

His question is about the bits/second rate, though.

The data is transferred as near enough to c that it really makes no difference to the end user, but it's the ability of both ends of the connection to process the data that's what actually matters in downloading things.

[TheBiggerFish]

Eh.

[Execute/Dumbo.exe]

That makes sense, 'the weakest link' and all that, thanks for the help, just remember it ain't a competition.

[Bauglir]

More importantly, the speed of data transfer depends on the speed of a symbol transmitted over the wire, moreso than the speed of the medium used to transmit it (at least with a medium like light over the distances we transmit). That is, how long the transmitter spends on each bit (or cluster of bits - if you've got, say, 4 different symbols, you could encode two bits into each one). A symbol is some kind of signal that symbolizes something, for instance a sequence of bits - one easy example is, using an electrical wire, setting the voltage to, say, 1 V to symbolize a 1 and -1 V to symbolize a 0. You have to spend enough time on each one that it's unambiguous which signal is being transmitted by the time it gets to the receiver, and it turns out that there are physical laws that constrain that, no matter how good your equipment gets.

here's a good book that covers a lot of that networking stuff, although it covers a lot more than the physical medium, if you're interested

[Starver]

I would have gone into the difference between latency (the time it takes data to get from point A to point B) and bandwidth (the amount of data that can be shuffled along).

The first is slightly complicated by the need to 'handshake': "Hey, I have data, can you receive?" "Yes, I can receive, what is your data?" "My data is <DATA>, did you get that?" "Yes, I got that data" "I have more data, can you receive?" ...etc, to a greater or lesser degree, depending on protocol, which means that there's a bit of thinking time and the need for messages (albeit often far shorter than the full data) to transit multiple times up and down the same fibre/copper-wire/radio-link/whatever, plus a bit of 'thinking time' at each end while the message is composed, plus possibly delays when the message is "No, I can't receive right now, hold on a moment because I'm busy" or "Sorry, I didn't get a response, just to repeat: <Can you receive?>/<What is your data?>/<Did you get that?>/<etc>".  (Exactly how much handshaking you need depends on a number of things, including which protocol layer of communication you're talking about...  Just assume that it exists, though.)

And if the data is in a large chunk, you can't do anything (even acknowledge, at one or other layer of protocol) with the chunk until it all arrives, relying more on the bandwidth.  If it's split into small chunks so for your end to re-assemble then you can acknowledge the small chunks (and re-request any small chunks that seem to be missing, for one reason or other), but you still won't be able to do anything with them if they need to be reassembled to have any usefulness, and there's a lot more "Tell me if you got that..." going on, especially if you didn't for some reason.

edit: And then I forgot to actually explain bandwidth, as opposed to the latency.  Never mind, that's addressed via the following two paragraphs...

To use a different analogy with less jargon to it, using semaphore or an Aldiss lamp transfers information literally at the speed of light, but is limited to the capabilities of both the operator sending and the one receiving.  And on a foggy day you might need to use bigger flags, brighter lights or even a chain of operators passing on the message between the source and destination, all of which slows down the effective transfer of information, even if it's passing at (pretty much) the speed of light over a majority of its distance.

Or to approach the question from completely the other direction, see this explanation...

[Helgoland]

Does anyone have a good source on the use of alkyne bonds in drug design? I remember reading somewhere that including those was some kind of fad a while ago, but I'd rather have something more substantial.

[Bumber]

So, my brother has Fiberoptic cable in his house, and he's said that his download speed can get up to 2-3 megabits per second (We live in Australia, alright?) but then asked me why its download speed is still weirdly slow, because if it uses light to transmit data, then the data should be transferred at light speed, right? And now I'm confused about that, too, why is it so slow if it uses something so fast to transmit data?

The primary cause of slowness would probably be mostly routing. After you request data from the server, the server sends packets to its local router. The router stores the packets, does error checking, and only once they're fully assembled can it start to send them off to the next router. If the next router is too full, the packets are lost and have to be re-transmitted from the previous router on a different route. You can expect at least 10+ different router hops if you're downloading something overseas. I think that also implies wireless transmissions, which can result in greater packet loss, as well as having lower capacity for simultaneous transmissions.

If you're curious, you can open up your command prompt and type "tracert <website>" and see what route it's taking and what the delay is between each hop.

[wierd]

Further reading related to routing delays--

https://en.wikipedia.org/wiki/Bufferbloat

[wierd]

Scientists at ATLAS detector may have detected new cousin of Higgs particle.

Since the standard model does not predict multiple Higgs particles, this may be new physics, and thus very exciting.

[Dutrius]

Oh cool, new phy...

url=http://www.foxnews.com/science/2015/12/16/science-breakthrough-physicists-may-have-discovered-higgs-boson-relative.html

Hmm... Let's see what CERN has to say.

However, the excess is too small at this stage to draw such a conclusion. We will have to wait for more data in 2016 to find out whether this slight excess is an inconsequential statistical fluctuation or, alternatively, a sign of the existence of a new phenomenon.

[wierd]

wiggleword: may.

insufficient data at this time. Yes, faux news. I know.

I hope it is more than just a statistical anomaly-- new physics is a very exciting possibility, but I wont get m hopes up just yet.

[Arx]

I also looked at it, thought 'I thought Wierd was big on reputable sourcing', and then had a look at the Nature article, which says much the same thing but differently. It's notable that two different rigs noticed a similar anomaly, though.