Database Reference
In-Depth Information
perform routing and consolidation functions in addition to the large array of data-
base functions. This option is slightly less expensive because we have to duplicate
routing and consolidation functionality only on the servers, not on the numerous
clients.
Architecture with Middleware (Three Tier)
This is a practical and efficient
approach to relieve the server and the clients of the additional routing and
consolidation functionality. You create a middle tier between the server and
client tiers. You install software, called middleware, with routing and consolidation
capability in the middle tier. Let us review how this architecture works for distrib-
uted database systems. At each site you have a database server machine, a middle-
ware machine, and as many client machines as needed. The server on one tier,
the clients on the other tier, and the middleware on the third tier form this three-
tier architecture. The server, middleware, and the clients are usually linked on a
LAN.
The following briefly describes how a three-tier system may apply to distributed
databases.
•
The server performs all database functions. Based on query requests, the server
retrieves the requested data and passes them on to the middle-tier. It also
updates the database as required by executing transactions.
•
The client manages the user interface, interprets queries and transactions, and
routes them to the middle tier. When data are received from the middle tier,
the client presents the information to the user.
•
The middle tier examines each query and routes it to the appropriate server.
If a query requires data from multiple sites, the middle tier breaks up the query
into appropriate subqueries and routes each subquery to the proper server.
When results are received from multiple servers, the middle tier consolidates
the result sets and sends the consolidated result to the requesting client.
The three-tier architecture proves to be a straightforward and efficient option.
The machines at each tier can be configured correctly for the functions they intend
to support. In practice, the systems tend to be multitier systems with other tiers such
as application servers added to the mix.
Design and Implementation Issues
By now, you are proficient at the design and implementation process for centralized
database systems. You are quite familiar with the phases of the database develop-
ment life cycle. After the initial planning and feasibility study phases, you move into
the requirements definition phase. For developing distributed database systems, you
do the same thing. The requirements definition phase forms the basis for the design
and implementation phases.
In the requirements phase, you gather information requirements for performing
logical and physical design. While designing a centralized database, you know that
the entire database will reside in one location, perhaps on a single database server.
This, however, is not the case with a distributed database. You do not store all the
Search WWH ::
Custom Search