馃攳
Network Throughput | Signal to Noise ratio | SNR | bandwidth vs throughput - YouTube
Channel: ISO Training Institute
[0]
I'll begin by introducing some fundamental data聽
units talk about the definition of through play聽聽
[12]
explain the concept of maximum throughput discuss聽
the difference between throughput and bandwidth聽聽
[19]
and talk about channel capacity a throughput can聽
be measured at any layer of the system so it can聽聽
[27]
be measured at any of the OSI model layers聽
physical up through application in order to聽聽
[36]
measure throughput we first need to look at some聽
fundamental data units now inside a computer all聽聽
[43]
data are represented as collections of digital聽
switches which are either on or off 1 or 0 an聽聽
[49]
individual digital switch is called a bit and in聽
networking this fundamental unit is represented by聽聽
[58]
a lowercase B according to the I Triple E 1541聽
standard or by the word bit written out with a聽聽
[66]
lowercase B using the IEC 80000-13 standard聽
a collection of 8 bits is called a byte and聽聽
[77]
that usage is standardized by IEC 80000-13 we say聽
that a byte always has a bits it is the smallest聽聽
[89]
addressable unit of storage inside most computers聽
most computers cannot address their individual聽聽
[96]
bits they can only address individual bytes and聽
sometimes they can't even address individual bytes聽聽
[102]
and dundas even larger than that but in networking聽
bites are represented by an uppercase B and this聽聽
[110]
is true for both the III EC 80000-13 and I Triple聽
E 1541 standards there is another term however聽聽
[122]
that we see in networking and this term originated聽
from the fact that certain early computers in the聽聽
[128]
1950s defined bites to have fewer than 8 bits thus聽
early data communications engineers use the term聽聽
[137]
octet to refer unambiguously to groups of 8 bits聽
in modern standard usage the word byte and the聽聽
[146]
word octet mean exactly the same thing but network聽
engineers typically use the term octet and you'll聽聽
[153]
see octet written in many networking texts and聽
networking documents the symbol for octet is a聽聽
[161]
lowercase o following the IEC 80000-13 standard聽
aside from networking texts lowercase o in place聽聽
[172]
of uppercase B in other words octet in place of聽
byte is used in non-english speaking countries聽聽
[179]
and in non-english texts especially French texts聽
so you'll often see things like one mo or one Gio聽聽
[188]
and that's simply the equivalent of one megabyte聽
or one gigabyte it's just mega octet or Giga聽聽
[195]
octet but they mean that they mean the same thing聽
because octet sand bytes are equivalent in modern聽聽
[200]
usage so by modern standards one octet is equal聽
to one byte and that's equivalent to eight bits聽聽
[209]
if you see a French text that says 1 capital M聽
lowercase o that is equivalent to one uppercase聽聽
[217]
M uppercase B which is equivalent to 1 million聽
octet or 1 million bytes or 8 million bits so聽聽
[230]
now that we have the fundamental unit of data聽
we can measure a three point throughput is a聽聽
[235]
very simple equation throughput is equal to the聽
amount of data that we transmit per unit time聽聽
[240]
so throughput is equal to data transmitted the聽
quantity of data transmitted divided by whatever聽聽
[247]
time unit we're using the standard unit that聽
we use for the numerator of this equation so聽聽
[255]
the standard unit that we use for the amount of聽
data transmitted is the bit and the standard unit聽聽
[262]
that we use for the amount amount of elapsed聽
time is the second so we normally measure聽聽
[269]
throughput in bits per second and we can use SI聽
or IEC prefixes in order to in order to give us聽聽
[277]
a convenient shorthand for units of thousands聽
or millions and so forth bits per second to PPS
[285]
now the maximum throughput of a networking聽
system is the maximum rate at which data聽聽
[292]
can be transmitted through a network connection聽
at whichever layer of the OSI model that we're聽聽
[297]
measuring so if we're measuring the physical聽
layer the maximum throughput is the maximum聽聽
[304]
rate at which we can send data through a network聽
connection at our using that particular physical聽聽
[311]
connection using that particular physical medium聽
but since each layer of the OSI model adds a聽聽
[318]
certain amount of protocol overhead in other words聽
it transmits some extra data added on to the data聽聽
[325]
that we actually would like to send for a given聽
maximum throughput at a lower layer the expected聽聽
[331]
maximum throughput as we're going to see it will聽
be less at the layer above it so there is a term聽聽
[342]
that is also used to describe maximum throughput聽
and that term is digital bandwidth it's measured聽聽
[349]
in bits per second measured at any layer in the聽
OSI model and it's kind of a weasel word because聽聽
[358]
the quote unquote bandwidth meaning digital聽
bandwidth of a networking system could actually聽聽
[365]
be set by a provisioning policy at the data link聽
network or transport layer and it might not be a聽聽
[371]
physical layer of the actual transmission sorry聽
a physical limitation of the actual transmission聽聽
[376]
system in order to avoid confusion though I'm聽
not going to use the term digital bandwidth聽聽
[383]
and instead gonna say maximum third byte and in my聽
lectures the word bandwidth is always referring to聽聽
[390]
a physical layer property and it is always a unit聽
that is expressed in Hertz I'm using the analog聽聽
[396]
signal processing definition of bandwidth so at聽
the physical layer the bandwidth capacity of the聽聽
[407]
physical medium does limit the maximum throughput聽
that's physically possible in other words at the聽聽
[413]
physical layer the amount of analog bandwidth I聽
have does play a role in determining the maximum聽聽
[422]
amount of throughput that I can transmit through聽
that particular physical medium in other words聽聽
[428]
it gives me a theoretical maximum speed and this聽
theoretical maximum speed is called the channel聽聽
[436]
channel capacity of the medium and that channel聽
capacity is given by this equation this is the聽聽
[444]
Shannon Hartley theorem and this says that the聽
channel capacity of a physical medium is equal聽聽
[451]
to its bandwidth in Hertz times the log of 1 plus聽
the ratio of the signal power expressed in watts聽聽
[460]
divided by the noise power expressed in Lights now聽
what I can see immediately from this equation is聽聽
[468]
that channel capacity is directly proportional to聽
bandwidth so as I increase my bandwidth my channel聽聽
[479]
capacity should increase assuming that my signal聽
and noise powers stay the same if I decrease the聽聽
[486]
bandwidth of a physical medium or I switch to a聽
physical medium that has a lower bandwidth the聽聽
[491]
channel capacity of that medium will be less than聽
my first medium assuming that my signal and noise聽聽
[498]
powers are the same but my signal and noise powers聽
do play a role in this bandwidth calculation for聽聽
[508]
example if I have an extremely high noise power聽
let's say I have a thousand watts of noise power聽聽
[517]
but I only have one watt of signal power this聽
one plus one over 1000 is still going to be聽聽
[526]
close to one and the log of one is zero which聽
means my channel capacity is going to be zero聽聽
[533]
there's too much noise for me to transmit any data聽
through that channel on the flip side if I could聽聽
[541]
somehow have a noiseless channel the limit as this聽
denominator approaches infinity of a finite signal聽聽
[550]
power over that noise power that's approaching聽
zero is infinity and infinity plus one is still聽聽
[557]
infinity the log of infinity is infinity times聽
whatever the bandwidth is I would have infinite聽聽
[563]
channel capacity so if I could have zero noise聽
and all signal my channel capacity could be聽聽
[569]
infinity it is a physical limitation given to聽
us by the laws of physics for any transmission聽聽
[577]
medium that we use that the noise power is not聽
going to be zero we are in fact going to create聽聽
[583]
a certain amount of noise simply by transmitting聽
the signal in the first place and the reason for聽聽
[589]
that is quite simple mess that the electrical聽
circuits that we're using transmit the signal聽聽
[595]
themselves introduce noise the lower the noise聽
produced by the electrical circuit more commonly聽聽
[602]
the more expensive the circuit so economics is聽
going to limit us to certain ranges for this聽聽
[609]
value and ultimately our channel capacity is going聽
to become much more dependent upon the bandwidth聽聽
[616]
so that channel capacity expresses the maximum聽
possible throughput of a physical link and it's聽聽
[624]
directly proportional to the amount of available聽
bandwidth it's also related to the signal-to-noise聽聽
[630]
ratio a good or high signal-to-noise ratio does聽
increase the capacity of the channel while a聽聽
[637]
bad or low signal-to-noise ratio decreases the聽
capacity of the channel and at a fixed signal聽聽
[644]
power high noise power is going to reduce the聽
maximum throughput available to the channel try聽聽
[650]
using a NATO 2.11 wireless network for example聽
in an apartment complex or block of flats British聽聽
[657]
English and you'll see that you'll never quite get聽
the speeds that is advertised by the particular聽聽
[665]
wireless networking standard and the reason for聽
that is that there's so much noise from everyone聽聽
[670]
else's wireless devices that it reduces the聽
maximum throughput that your wireless devices can聽聽
[678]
achieve because the channel is simply too noisy a聽
noiseless channel a completely noiseless channel聽聽
[685]
is physically impossible because modulation聽
transmission reception demodulation circuits all聽聽
[690]
add noise and even if we're sending data down a聽
cable or down a fiber optic line properties of the聽聽
[698]
materials themselves introduce a certain amount of聽
noise in the signal now the component quality does聽聽
[704]
affect the amount of noise that is introduced聽
higher quality components for example will聽聽
[710]
tend to introduce less noise but higher quality聽
components have a higher manufacturing cost and聽聽
[716]
so we have to consider the economic realities聽
of how much noise is acceptable for a given聽聽
[722]
location based upon the price of the components聽
that we would need in order to reduce that amount聽聽
[728]
of noise so to summarize the fundamental unit聽
of data that we use is the bit bites and octets聽聽
[736]
are groups of 8 bits in modern usage those two聽
units are equivalent to each other throughput is聽聽
[743]
the amount of data that we can transmit per聽
unit time and fundamentally we normally use聽聽
[749]
bits per second as our unit for expressing this聽
quantity the maximum throughput of the system is聽聽
[756]
the fastest data transmission speed that can聽
be achieved over a network link this maximum聽聽
[763]
throughput can be measured at any layer of the聽
OSI model but it is ultimately subject to hard聽聽
[770]
physical limitations that are introduced by the聽
physical channel capacity in the physical layer
[786]
you
Most Recent Videos:
You can go back to the homepage right here: Homepage