Datasheet

SQL Server 2000 – Particulars and History
19
2-Tier Architecture (Client-Server)
2-tier, or client-server systems, first started becoming popular in the early 90's. There were actually two
sub-types to this architecture: client-centric (smart client) and server-centric (smart server).
Client-Centric
The client-centric version of client-server was based on the notion that PCs are cheap (the driving force
behind most client-server development) and that you're going to get the most power when you
distribute the computing requirements as much as possible. As such, whenever possible, only the data
services part was performed on the server. The business and UI side of things was performed at the
client – thus ensuring that no one system had to do all that much of the work. Every computer did its
fair share (at least, that was the idea).
Traditional Client/Server
(Client-Centric)
Client
User & Business
Services
Server
Data Services
Database
The big problem with client-centric client-server was (and still is) bandwidth. If all the business logic is on the
client, then there tends to be a very large number of round trips (network sends and receives) between the
client and the server. Frequently, large chunks of raw data are sent to the client – quickly clogging the
network and slowing down everyone else trying to get their own huge blocks of data back and forth.
Advantages Disadvantages
Distributes the workload to a large number
of relatively cheap clients.
Is a terrible bandwidth hog – clogs networks up
very quickly.
If you have one user who needs more
speed, you can purchase a faster system for
just them rather than a large expensive host
system that everyone is going to take a
piece of.
Installations are time consuming and difficult to
coordinate. New software or versions of software
must be installed on multiple machines. Version
upgrades can be particularly problematic since
old clients are not always compatible with the
new server components and vice versa. All clients
may have to be upgraded at one time, which can
create quite a serious logistics problem.
The same money that buys the computing
power on the client side also buys power for
other productivity applications, such as word
processing and spreadsheet applications.
Each client, depending on the vendor, may need
a separate license for each seat or connection.
This can increase costs.