Server (computing)

A server is a computer program or a process that provides data, resources, or services to other computers. These other computers are known as clients. This specific architecture is called the client–server model.

In this system, a single server can serve multiple clients at once. Conversely, one client can connect to and use multiple different servers. While the term often refers to a physical device, it strictly refers to the software process running on that device. Any general-purpose computer can host a server if it runs the right software. This means a laptop can act as a web server just as easily as a massive machine.

Most client–server systems operate using the request–response model. This is a specific sequence of communication between the two parties. First, a client sends a request to the server for a specific action or piece of data. The server then performs that action and sends a response back to the client. This response typically contains a result or an acknowledgment of the request.

There is another method called the publish–subscribe pattern. In this pattern, clients register with a server to subscribe to specific types of messages. Once registered, the server pushes matching messages to the clients without needing a new request each time.

Servers are categorized by the specific tasks they perform for their clients. A web server hosts web pages, which is what makes the World Wide Web possible. Mail servers manage email communication, acting much like a digital post office. Database servers maintain and share organized collections of data, such as spreadsheets or accounting information. File servers share folders and storage space across a network.

Other specialized types include game servers for multiplayer video games and media servers for streaming audio or video content. Some servers, called proxy servers, act as intermediaries to control traffic or improve performance.

The history of the term "server" is rooted in queueing theory from the mid-20th century. In the early days of computing, people used more concrete terms like "telephone operators." By 1969, the term appeared in RFC 5, a document describing ARPANET, the predecessor to the modern Internet. That document distinguished between a "server-host" and a "user-host." This helped clarify the difference between the machine providing the service and the person using it.

Over time, the definition evolved to describe any process performing a service for remote requests.

Because servers are essential, they often require specialized "server-class hardware." These machines are typically more powerful and reliable than standard personal computers. Large enterprise servers are designed for high availability, meaning they must run for long periods without interruption. To prevent failure, they often use hardware redundancy. This includes dual power supplies and RAID disk systems for data protection. They may also use ECC memory to prevent errors.

Some components are "hot swappable," which allows technicians to replace parts while the server is still running.

Many large-scale servers are housed in dedicated data centers. These facilities provide a controlled environment with stable power and high-speed Internet. Because these machines generate significant heat, data centers use advanced air conditioning. Some high-end servers even use water cooling to manage temperatures.

Many servers are also designed to be rack-mounted. They are flat and wide so they can fit into standard 19-inch racks. These servers often run unattended without a monitor, keyboard, or mouse. Instead, administrators manage them remotely using tools like SSH or web-based management systems.

The client–server model is the foundation of almost the entire Internet. Millions of servers are connected globally, running continuously to handle user actions. High-level systems like DNS and routers direct traffic to the correct destinations. While most of the Internet relies on this model, there are exceptions. For example, peer-to-peer file sharing uses a different relationship where computers interact directly without a dedicated central server.