World Wide web based on 3 tier architecture, not client server?

[PdT]

I have to answer the following question, could someone please give me advice on this?

"The world wide web is implemented based on a 3-tier architecture or N-tier architecture; whereas the Internet architecture is implemented based on a client-server architecture. Please explain why the world wide web cannot be implemented based on a client-server architecture. "

[tao muon]

The terms "client" and "server" are quite ambiguous.

If you are looking for information, you are a client. If you have the information, you are a server.

So, looking for something makes you the client, and having it makes you a server.

But if the server your client first communicates with doesn't have the information, then it (the server) has to have a way to find it- thus creating the DNS system.

You have just transitioned from a pure client-server system to an N-tier system. DNS provides a way for a server to escalate and translate server location 'seemlessly' across the network.

And now the CS majors will come in and correct me.

[markwolfe Veteran]

I'm not an expert, but in client-server, the server really has a lot of work to do with modern content that often relies on processing information from a database and such.

With multi-tiered, there is more than one server, with functions divided up: for example, the database backend is running on one server, and the content (including results of database queries) is delivered generally through what I guess would be the main server for the application (and there can be several of these working together, but doing the same function, really, so I don't think that alone would count as multi-tier).

The benefit of multi-tier is that is very scalable. Add in more database servers if you have very heavy database requirements, and so forth.

As far as why the www "cannot" be implemented as a pure client-server, I guess that it really could, but the dynamic content would have to be processed by the same machines handling all the incoming requests. That could bring a popular site/server to its knees quickly. I don't think it would be practical.

I am sure that others with degrees in this sort of thing can give you a much better answer shortly.

EDIT: tao already beat me to the post, and he considers also the DNS servers, which I neglected.