DATA INTEGRITY - Database Design and Development: An Essential Guide for IT Professionals - page 444

Output phase Provide messages to be passed along regarding the

successful or unsuccessful completion of the database
interaction. Retrieve and provide data for completion
of the application program. or update database as
needed.

Transaction Types From the point of view of database interactions, transactions
fall into two broad types:

Read-only transactions. These do not change the database content. Therefore,
any number of read-only transactions can share the database content
simultaneously.

Update transactions. These change the database content. Therefore, while pro-
cessing update transactions, conflicts could arise among simultaneous transactions
attempting to update the same data items.

Single-User Environment Personal database systems are single-user environ-
ments. Concurrent transactions and resulting concurrency problems are typically
absent in such an environment. However, different types of hardware and software
failures could cause data integrity problems.

Multi-User Environment Most database systems are multiuser environments. In
such environments, concurrent transactions are the norm. Concurrent transactions
could give rise to transaction conflicts.

Concurrency Control A multiuser environment requires effective measures and
techniques for concurrency control. Each transaction must perform independently
as though it were executing in isolation. Concurrent transactions must employ
locking mechanisms to prohibit other transactions attempting to update the same
data items.

Data Granularity A database is a collection of named data items intended to be
shared among multiple users. When multiple users need to update the same item,
the data item must be locked by one transaction while it goes through the database
interaction phase. How many data items do you lock and prevent other users from
updating? Just specific fields, specified rows, or the whole table? Data granularity,
in this context, refers to the level and scope of database content that needs to be
locked at a given time.

Dependability When user requests are processed as transactions, the users expect
the transactions to preserve the integrity of the database. Processing of concurrent
transactions in parallel must be transparent to the users.

Availability The database must be kept available to the users as and when needed.
Even after system and hardware failures, the database must be quickly restored to
a stable and consistent state.