Question on Distributed Database Management System
Date: 20 Jul 2001 01:50:32 -0700
Message-ID: <52e38344.0107200050.6447c80c_at_posting.google.com>
Dear all,
I have recently been considering a database system for my company. The problem is that my company has a number of sister companies. Company A maybe running a DBMS A, Company B running DBMS B, and Company C runnning DBMS C. I would of course like to integrate all of them. Digging back to my database design book, it seems like a distributed DBMS, in particular the federated multidatabase system (FMDBS), is the way to go.
My reason for thinking in terms of FMDBS is because each company operates their own business and therefore should have complete autonomy to their data. They should be the one to decide what information to give out and what information to remain confidential. They should also have the flexibitily to upgrade and maintain their own system. The principle of FMDBS seems to be the best concept to grasp to allow local autonomy.
However, being very new to the concept of DDBMS, I then got a bit lost...
- If DBMS A, B and C are all different type of database system (for example, Oracle, Sybase, mySQL), building a web based interface that queries directly using SQL statements, would this be considered a DDBMS? And since I don't touch the underlying structure of the local database, would this sort of solution be classified as a FMDBS?
- It seems that a unified DBMS would be a easier approach to integrate all the data within different companies. The advantage I assume for FMDBS is that it's a cheaper solution or a solution that does not interfere greatly with the existing architecture. Is this a correct observation?
- Since each company carry their own data, for example company A has a list of customer's address that may overlap with company B's customers, it seems that the only solution is to have a central database to prevent overlapping. Since company A and B has privacy issues, these data cannot be shared in a central database server. This seems to go against all rules of "Correctness of fragmentation". How does one design with these obsticles.
My sincere thanks for any comments you may have on these.
Regards.
Kelvin Received on Fri Jul 20 2001 - 10:50:32 CEST