Network Engineering Stack Exchange is a question and answer site for network engineers. It only takes a minute to sign up.
Sign up to join this community
I totally understand the difference between bandwidth and throughput. While bandwidth shows the maximum amount of data can be transmitted from a sender to a receiver, throughput is the actual amount of data that has been transmitted as they could be different factors such as latency affecting throughput.
Bit rate is the amount of data(number of bits) can be transmitted per second which sounds the same as throughput to me. So what is the key difference?
Some of these terms are used differently by different people, but below is what is generally accepted.
Bandwidth is the number of bits per second that a link can send or receive, including all flows. For example, the bandwidth of a 100 Mbps connections is 100 Mbps, but that doesn't mean it is always sending or receiving 100 Mbps, but that is the maximum possible on that link. Unlike what many people mean by bandwidth, it does not mean data usage. I see people say that they have a bandwidth limitation (every link does), and they have used all their bandwidth for the month. This is an incorrect use of the term. What they mean to say is that they have a data usage limitation, and they have used it up for the month.
Throughput is the amount of data during a time period that a flow (process to process) can send or receive. This includes all the host overhead, and contention on the link (multiple flows on a link will each use some percentage of the bandwidth, reducing the throughput of each).
Bit rate is closer to bandwidth, but it is often per host, or source to destination devices. You may have a bit rate of 100 Mbps from a host to a switch, but the bit rate from a host to a host is less. This usually includes multiple flows.
Daily I see peoples even specialists in communication do mistakes about the three mentioned terms:
Bandwidth: The unit of it is Hz, so it is mathematically is: High_Used_Frequency - Low_Used_Frequency. So, when we measure bandwidth in bps, i.e we do mistake. Besides, some guys working in Network field, totally they treat with the bandwidth as Data rate. So, bandwidth is the difference between high and low frequency. For example, bandwidth of a channel in FM is 0.2MHz, for a carrier is WCDMA is 5MHz.
Data Rate: The unit is bps, so we can say that the data rate of the channel is the maximum ability of the channel that can transmit bits in a second. In the normal case, we can say that if we have 5MHz bandwidth, then the channel capacity (or data rate) is 5Mbps, but also it can be that the bandwidth is 5MHz but the data rate is 10Mbps, or 2.5Mbps, this is depends on the modulation type and some other technics.
Throughput: This term is used in many different systems, shortly, and clearly we can define it as: the real data that we receive. For example, the Bandwidth=5MHz, channel capacity(data rate)=10Mbps,but what we receive exactly is 6Mpbs due to any reason, so the throughput is 6Mbps... Hope I gave you an idea...
Not disagreeing with other answers, but an important takeaway is that people often are meaning slightly different things with similar and identical words.
A couple of points which are worth adding, especially if you do anything numerical like try to work out the maximum throughput of a given FTP connection over an X Mbit/sec line:
If you ever get your oscilloscope on the wire you'll quickly find out that the symbol rate can be different. 100baseT for example is 125 Mbaud, because it transmits 5 bits on the wire for 4 bits of data.
In summary, the terms used often vary depending on what you're emphasising:
You sometimes will see things like "peak throughput" and "average sustained throughput" which I'd regard as very helpful. Very often the rates that are interesting depend entirely on the period you average over.
PS IEC prefixes kibi, mebi, etc are far from universally used but worth knowing about
I'll try to explain it very simply. BW is the range of frequencies that can pass through a channel ( Unit: Hz). Data rate is the amount of data that passes through a medium per unit time ( Unit: bits/sec). Imagine a narrow road and highway, road width as the BW and car going through as data. More BW, More data rate. So, data rate is data passing through medium. But if the data is maximum then it is equal to bandwidth. BW is the maximum data rate.
According to Shannon-Hartley law/theorem for noisy channel:
Data rate = bandwidth x LOG2(1 + signal-to-noise ratio)
The signal-to-noise ratio depends on bandwidth, signal power and channel noise power. Therefore, bandwidth is in general never the same as data rate. They are the same only in the special case where the signal-to-noise ratio equals 1.
Throughput is the effective or the actual data rate on a link, so it is at most equal to the data rate.
All of these answers are at least partially wrong but some contains partial truths. Interestingly, the wrongest one is the accepted one, which contains an unacceptable misunderstanding of what is the bandwidth. I am a postdoc and assistant professor in Telecommunications Engineering (I teach Telecommunications Networks), so I offer a technical view on the topic.
Bandwidth: B is the range of frequencies that is used for the communication and the unit of measure is Hz. Never confuse the bandwidth with the bitrate or the throughput. This is a common mistake but they really have different physical meanings (nothing to do with information!).
Bitrate: this is the number of bits that can be sent over a single telecommunications link in one second. Its unit of measure is therefore b/s or bps. We indicate it as Rb.
Throughput: this is the highest achievable bitrate from the source to the destination, over all the links. Of course, the unit of measure is the same of the bitrate. Here, we consider the bottleneck link so the throughput is min{Rb1,...,Rbn}. We can never go faster than the slowest link...
How are these quantities related to the Shannon-Hartley theorem? The Shannon-Hartley theorem defines the number of bits that you can theoretically send on the channel in a second. This is called the channel capacity C and computed as C = B * log2(1 + SNR) [bps], but this is not equal to Rb. Actually, we have a further theorem: Rb <= C. This is because Rb accounts for protocol-, hardware-, and software-dependent losses and it must meet a series of requirements. The channel capacity is just a theoretical limit. Commercial equipments set pre-defined discrete values of Rb mainly based on the value of the SNR.
I hope to have added... a useful bit of information!
