Portfolio
Distributed DBMS decisions: Will you go with a client/server DBMS or a true distributed DBMS?
Source: Computerworld Published: May 1991
In this 1991 article, George Schussel defined the difference between client/server DBMS and distributed DBMS. Some comments from the article are:
The difference between true distributed DBMSs and client/server DBMSs is in the concept of location transparency. With location transparency, a program running at any node need not know the physical location of the computer in which the requested data resides. True distributed DBMSs support location transparency, with each separate physical node in the network running a copy of the DBMS and associated data dictionary. It is the true distributed DBMS’ responsibility to determine an access strategy to that data.
Distributed DBMS software, of which client/server DBMSs and true distributed DBMSs are a part, has to provide all the functionality of multiuser mainframe database software and allow the data in the database to exist on a number of different but physically connected computers.
Distributed DBMSs should have the following functions:
- Data integrity through automatically locking records and rolling back partially complete transactions.
- The ability to attack deadlocks, automatically recovering completed transactions in the event of system failure.
- The ability to optimize data access for a wide variety of application demands.
- Specialized I/O handling and space management techniques to ensure fast and stable transaction throughput.
- Full database security and administration utilities.
Download the entire article for a comprehensive overview.
