Low latency is the lower time delay in a computer system or communication network. It is most often used in reference to the amount of time it takes a network packet to make a round trip from a sending computer to a receiving computer via a network connection. In other use cases, it may refer to amount of time it takes data to be directed into and out of a computer's random access memory (RAM).
Using a ping test is the most reliable method for working out the latency in a network communication. Calculating the latency on a computer network involves recording the time it takes a ping to reach its destination and return. This may vary depending on a number of variables, including the time of day, type of network, distance to the destination, and more. Typically, a number of pings is sent, and the latency is an average based on tallying each ping's round trip time.
Low latency is highly useful in a number of scenarios, but doesn't have anything to do with bulk transmission of data. A connection may only have to link up with a host once to download a large file, the speed of which doesn't depend on the time delay in communicating back and forth with the source. The more desirable scenarios for low latency, then, are where multiple, repeated communications with the destination are required.
In voice over Internet protocol (VoIP) communications, such low latency is beneficial since the voice data needs to be sent back and forth at extremely high rates in order to avoid delay in conversation. This is also true for certain types of networked applications such as massively multi-player on-line (MMO) games, and real-time trading in markets. In gaming, the client playing the game needs to be able to update his location within the game continuously throughout play. Higher latency creates a lag in the game's performance for the player, since the server hosting the game isn't able to update the player's position as quickly. Other network applications, such as those used by day-traders, find low latency desirable in being able to update and modify a trading portfolio as quickly as possible to keep up with some high-frequency markets.
Though most latency discussions revolve around computer networking, the specific hardware inside a computer can gain a benefit from low latency as well. Computer RAM is one example where the time it takes for data to be added to or removed from a memory column can improve performance. This is referred to as column address strobe (CAS) latency, where a device in the computer called the memory controller directs data into and out of specific columns in the RAM. Some memory controllers are on the computer's motherboard, while others have improved latency by implementing them on the central processing unit (CPU) instead.
Another computer hardware implementation that improves on latency is in some peripheral audio devices. Some manufacturers use special device drivers for their expansion cards that bypass the operating system. Audio software that is more compliant with these special device drivers then achieve a lower latency in delivering sound from a program through the device to the user's ears.