Structure of the Internet: Client server model
The history of computing started off with centralised computers (in many cases mainframes) or servers performing all the calculations. Client computers were then attached to these centralised computers (servers) and if you wanted to calculate something, you would have to wait for the central computer to respond. As computing power got cheaper client nodes became more powerful and the central computer less important. However, with the growth of the internet, there has been a shift back to a client server model. Powerful central computers store information such as emails, documents, music and videos or offer services such as file hosting, printing, game hosting and internet access; client computers fetch information and use services from these central servers. In the next few years you are likely to see more and more software moving away from running on your desktop to running on remote servers and you accessing it as a client, this is called software as a service.
As an example of modern client server model consider a video sharing website. The website, let's call it mutube, has a server that stores all the videos that are uploaded to the site. The website is used by millions of clients a day and each of them connects to the server to watch videos. When a client connects to mutube the server and asks for a particular video, the server loads the video into RAM from a large array of hard disks and mutube sends the video to the client. The client on receiving the video, presses play and watches the video.
Other examples of servers might be a shared printing service in a college. The print server will be hosted on a single computer, and when anyone in the college wants to print, the request is sent to the server. In this case the server will keep track of how much printing credit each user has and make sure that the print queue is dealt with properly.
The current client-server model is starting to change, with companies being less likely to offer services with a centralised server. Increasingly internet firms are reaching a global clientèle, it makes little sense to have a server or servers based in one location as if your servers are in America and some of your users in Armenia, these users will experience slow access to your services. Another problem is if a power cut affects your server or the connection to that one server or set of servers goes down then the service you are offering the internet will also stop.
With cloud computing the services may be distributed all over the globe, meaning that wherever you are, you'll have a server reasonably close to you offering access to the data and services you need. It also means that if one server goes down other servers in different locations can keep the service running. Keeping databases synchronised across the globe, so your mail client has the same mails in Switzerland as in Swaziland, is a complex task and firms such as amazon and rackspace offer services to help you handle this. One downside with cloud computing is you are never quite sure where your data is, and if you're not careful you might find data being stored in countries that have less stringent data protection laws than your own.
Servers are software programs that in most cases run off normal computing hardware. Server software includes:
- File sharing
- Game hosting
- Other web services
Clients are software programs and processes that connect to servers, sending requests and receiving responses. Client examples include:
- Web browser page requests
- Chat systems on mobile phones
- Online games
Serving websites, hosting games, file sharing, printer sharing