DISTRIBUTED DATABASE SYSTEMS - Database Design and Development: An Essential Guide for IT Professionals - page 558

New York London

Data Replication

DB DB

Employee
File -- Full

Employee
File -- Full

DB DB

Employee
File -
French Only

Employee
File -
Names Only

Data Fragmentation

Paris Hong Kong

Figure 18-9 Data fragmentation and data replication.

EMPLOYEE relation into subsets of rows relating to employees in these three
regions.
Figure 18-10 demonstrates horizontal fragmentation. Note how the subsets are
stored at the three sites.

Vertical fragmentation Each fragment consists of a subset of columns of a rela-
tion. Each subset must also contain the key column in addition to the other selected
columns. Sometimes, a tuple-ID is added to each tuple to uniquely identify it. How
can you use this method to create fragments from the EMPLOYEE relation? Let
us say that for your organization the human resources functions are localized in
Boston and the accounting functions, including payroll, are localized in New York.
When you examine the columns of the EMPLOYEE relation, you will note that
there are a few columns that are of interest only for payroll and the remaining
columns are desirable for the human resources division. Break up the EMPLOYEE
relation into two subsets of columns to be stored at the two sites.
Figure 18-11 demonstrates vertical fragmentation. Note how the subsets are
stored at the two sites.

Advantages and disadvantages Fragmentation improves efficiency in data access.
Data fragments are stored at the sites where they are used most. Local query and
transaction performance gets better. At the same time, you get enhanced data
security because it is easier to protect the local data.
On the other hand, when any site needs data from other sites for any types of
queries or transactions, access speeds become inconsistent. Queries on global data