Thin- versus thick-client computing

Overview

Recall questions on where the processing sits in each model, the hidden costs of cheap client hardware, and which model survives a network failure.

What you need to know

Answer each one before you reveal it.

Questions from this subtopic

Each one with the model answer underneath.

1. 1

   What is thin-client computing?

   Model answer

   In thin-client computing almost all of the processing and storage happens on a powerful central server. The client device does little more than accept input, send it to the server, and display the results the server sends back. It needs only modest hardware and holds little or no data or application software of its own.

2. 2

   What is thick-client computing?

   Model answer

   In thick-client computing the client machine does most of the processing itself and holds its own applications and, usually, its own data. The server's role is reduced to supplying shared files or data when needed, or may be absent altogether, so the client can do useful work independently.

3. 3

   Compare and contrast thin-client and thick-client computing.

   Model answer

   The difference is where the processing and storage sit. Thin-client puts them on the server: client hardware is cheap and long-lived, software is installed and updated once centrally rather than on every machine, data is held and backed up in one place, and security is easier to enforce because little is stored on the client. But it depends utterly on the network and the server — if either fails, no client can do anything — the server must be powerful and is expensive, and as more clients are added the load on server and network rises, which can slow everyone down. Thick-client puts them on the client: it keeps working when the network is down, gives faster response for demanding tasks because the work is done locally with no round trip, and spreads the load so adding a client adds its own processing power. But each client needs capable and therefore expensive hardware, software must be installed and updated on every machine individually, data scattered across machines is harder to back up and secure, and a lost or stolen machine may take data with it. Both are alike in following the client server model of requests and responses; they differ only in how much of the work travels across the network.

4. 4

   Give two situations where a thin-client approach is the better choice.

   Model answer

   Any two of: a school or library with large numbers of machines doing light, similar work, where cheap terminals and central management save a great deal of cost and administration; a call centre or booking office where every user runs the same one application against a central database; an environment where data must not leave the premises or be stored on individual machines for security reasons; and any organisation where software must be updated frequently across many machines at once.

5. 5

   Give two situations where a thick-client approach is the better choice.

   Model answer

   Any two of: work that is computationally heavy or graphically demanding, such as video editing, 3D rendering, engineering software or gaming, where sending everything to a server and back would be far too slow; users who must be able to work while disconnected, such as staff working in the field or travelling; a small organisation that cannot justify the cost of a high-specification server; and situations where each user runs quite different software from the others, so central provisioning brings no benefit.

6. 6

   Why can the client hardware be cheaper in a thin-client system, and what is the hidden cost?

   Model answer

   The client only has to handle input and display output, so it needs little processing power, little memory and little or no storage, and such a machine is cheap to buy and stays usable for years because rising software demands fall on the server rather than on it. The hidden cost is that the server must now be powerful enough to do the work of every client at once, and the network must be fast and reliable enough to carry all of that traffic; both are expensive, and both must be provisioned for the busiest moment.

Other ways to work this

The same content, four other ways.

What is inside Levo

Six ways to work the same material.