Databases Reference
In-Depth Information
sive amount of processing traffic on a server can cause monumental per-
formance problems in the distributed database community.
The other issue for a server node is its storage capacity. This is critical
because the server maintains the central repository for data in the distrib-
uted database community. This central repository may be concentrated
locally in this one server node or it may be spread across several other
remote server nodes.
As compared with the server node, the client node can be limited in its
amount of processing power and storage capacity. Client nodes are typi-
cally smaller desktop microcomputers or workstations. The exception to
this rule occurs when a node acts as both a client and a server. Other issues
that can dictate the amount of processing power and storage capacity on
the client node are the amount of data redundancy kept on the node and
the storage requirements for its application software.
SOFTWARE DESIGN
For most commercially available distributed DBMSs, the software con-
sists of a family of products that are available on a variety of hardware plat-
forms. A typical family of products might include:
• The basic DBMS and its active data dictionary.
• The communications software that is coupled with the DBMS .This
software may be available with various levels of capability. For exam-
ple, the minimal capability would be a protocol for remote data ac-
cess. The next level of capability would be a gateway for remotely
accessing foreign databases or files. (Foreign databases are databases
established by other brands of DBMS software.) The truly distributed
functional capability would be a communications software product
that supports location transparency data accesses and concurrence
control. It would also include such features as a two-phase commit
protocol for ensuring data consistency.
• Some distributed DBMS vendors also offer additional software utilities
(e.g., fourth-generation languages, Query-By-Forms or Query-By-Ex-
ample, fancy report writers, and DBA tools for monitoring activities).
One of the first decisions the organization must resolve is the make-or-
buy decision. To be practical, unless the application is so unique that none
of the commercially available distributed DBMSs will suit the needs, it is
advisable not to build a home-grown distributed DBMS. Once the organiza-
tion decides to buy, a survey and feature analysis of the market must be
performed. The selection criteria must also take into consideration the
amount and the types of software packages that will also be operating
within the same platform.
Search WWH ::
Custom Search