Network latency is an expression of how much time it takes for a packet of data to get from one designated point to another. In some environments (for example, AT&T), latency is measured by sending a packet that is returned to the sender; the round-trip time is considered the latency. Ideally latency is as close to zero as possible.

Packet delay within the communication subnet (i.e., the network layer). This delay is the sum of delays on each subnet link traversed by the packet. Each link delay in turn consists of four components.

1-The processing   delay between the time the packet is correctly received at the head node of the link and the time the packet is assigned to an outgoing link queue for transmission. (In some systems, we must add to this delay some additional processing time at the DLC and physical layers.)

2. The queuing   delay between the time the packet is assigned to a queue for transmission and the time it starts being transmitted. During this time, the packet waits while other packets in the transmission queue are transmitted

3. The transmission delay between the times that the first and last bits of the packet are transmitted

4. The propagation delay between the time the last bit is transmitted at the head node of the link and the time the last bit is received at the tail node. This is proportional to the physical distance between transmitter and receiver; it can be relatively substantial, particularly for a satellite link or a very high speed link.

Latency from Wikipedia:

Latency is a time delay between the moment something is initiated, and the moment one of its effects begins or becomes detectable. The word derives from the fact that during the period of latency the effects of an action are latent, meaning “potential” or “not yet observed”.

Most people understand that it takes time for web pages to load and for emails to get from your outbox to the destination inbox and yes, this is a form of latency. But in order to understand why this happens we need to think about latency at a lower level: Latency is a time delay imparted by each element involved in the transmission of data.

according to RFC 3393

The variation in packet delay is sometimes called "jitter". This term, however, causes confusion because it is used in different ways by different groups of people.

"Jitter" commonly has two meanings: The first meaning is the variation of a signal with respect to some clock signal, where the arrival time of the signal is expected to coincide with the arrival of the clock signal. This meaning is used with reference to synchronous signals and might be used to measure the quality of circuit emulation, for example. There is also a metric called "wander" used in this context.

The second meaning has to do with the variation of a metric (e.g., delay) with respect to some reference metric (e.g., average delay or minimum delay). This meaning is frequently used by computer scientists and frequently (but not always) refers to variation in delay.

Network latency is an expression of how much time it takes for a packet of data to get from one designated point to another. In some environments (for example, AT&T), latency is measured by sending a packet that is returned to the sender; the round-trip time is considered the latency. Ideally latency is as close to zero as possible.

Packet delay within the communication subnet (i.e., the network layer). This delay is the sum of delays on each subnet link traversed by the packet. Each link delay in turn consists of four components.

1-The processing   delay between the time the packet is correctly received at the head node of the link and the time the packet is assigned to an outgoing link queue for transmission. (In some systems, we must add to this delay some additional processing time at the DLC and physical layers.)

2. The queuing   delay between the time the packet is assigned to a queue for transmission and the time it starts being transmitted. During this time, the packet waits while other packets in the transmission queue are transmitted

3. The transmission delay between the times that the first and last bits of the packet are transmitted

4. The propagation delay between the time the last bit is transmitted at the head node of the link and the time the last bit is received at the tail node. This is proportional to the physical distance between transmitter and receiver; it can be relatively substantial, particularly for a satellite link or a very high speed link.

Informally at least, delay and network latency are the same thing: the time that it takes a packet to get from point A to point B.

For lots of reasons, that's not a single number, that's really a probability distribution function, with a lower bound given by the laws of physics.

Ideally, you'd have a pure spike at this lower bound, but practical reasons turn that into some other kind of curve. As in statistics, the variance is simply telling you about how 'wide' or 'broad' that curve is, and is trying to capture how much change there is in the network latency.

The network latency is specified from the start of the cell being transmitted at the source to the start of the cell being received at the destination

While the variance of delay is of strong interest to real-time multimedia applications such as voice and video streaming , it  is  largely due to transient congestion caused by traffic variability, routing instability, and network failures.

You have some excellent answers already, so all I will do is discuss how ATM might differ, in this regard, from Ethernet.

As you know, ATM can be adjusted for class and quality of service. So perhaps we take something obvious, to illustrate the differences best. Take an ATM network, where communications are constant bit rate (CBR), each user having a guaranteed and rather restricted bit rate. The bit rate has to be low enough that every user can expect that level of service at all times.

Compare this with a switched Ethernet catenet. A device has frames to transmit, and sends them out at wire speed. The fastest speed available.

In practice, Ethernets are overprovisioned to a fair degree. This means that the frames transmitted by a device will often go through this catenet very quickly, encountering mostly very short queues at the merge points. But there will be those statistically inevitable times, where the queues at merge points are instead long.

So with the Ethernet, you can expect a long-tailed delay distribution, something like Poisson distributed. Very short delays most of the time, because the wire speed transmissions go straight through to destination, and in less lucky times, delays caused by congestion. Instead, in the CBR ATM example, you will see more average latency, caused by the lowish bit rate for each virtual circuit, but you won't notice that long tailed delay variation. The ATM net was configured specifically to "guarantee" that CBR for every user.

Thanks Albert Manfredi  for your useful information,,  

Network latency is an expression of how much time it takes for a packet of data to get from one designated point to another. In some environments (for example, AT&T), latency is measured by sending a packet that is returned to the sender; the round-trip time is considered the latency. 

On the other hand, The variance of delay is of strong interest to real-time multimedia applications such as voice and video streaming , it  is  largely due to transient congestion caused by traffic variability, routing instability, and network failures.

Latency is the delay from input into a system to desired outcome; the term is understood slightly differently in various contexts and latency issues also vary from one system to another. 

Latency greatly affects how usable and enjoyable electronic and mechanical devices as well as communications are.

Latency in communication is demonstrated in live transmissions from various points on the earth as the communication hops between a ground transmitter and a satellite and from a satellite to a receiver each take time. People connecting from distances to these live events can be seen to have to wait for responses. This latency is the wait time introduced by the signal travelling the geographical distance as well as over the various pieces of communications equipment. Even fiber optics are limited by more than just the speed of light, as the refractive index of the cable and all repeaters or amplifiers along their length introduce delays. 

Network delay is an important design and performance characteristic of a computer network or telecommunications network. The delay of a network specifies how long it takes for a bit of data to travel across the network from one node or endpoint to another. It is typically measured in multiples or fractions of seconds. Delay may differ slightly, depending on the location of the specific pair of communicating nodes. Although users only care about the total delay of a network,[citation needed] engineers need to perform precise measurements. Thus, engineers usually report both the maximum and average delay, and they divide the delay into several parts:

Processing delay – time routers take to process the packet header

Queuing delay – time the packet spends in routing queues

Transmission delay – time it takes to push the packet's bits onto the link

Propagation delay – time for a signal to reach its destination

There is a certain minimum level of delay that will be experienced due to the time it takes to transmit a packet serially through a link. Onto this is added a more variable level of delay due to network congestion. IP network delays can range from just a few milliseconds to several hundred milliseconds.