Databases Reference
In-Depth Information
data model:
An organization of data that describes the relationships among the primi-
tive and composite data elements.
data partitioning:
See partitioning.
data warehouse:
A large repository of historical data that can be integrated for decision
support.
DBA:
See database administrator.
denormalization:
The consolidation of database tables to increase performance in data
retrieval (query), despite the potential loss of data integrity. Decisions on when to
denormalize tables are based on cost/benefit analysis by the DBA.
dense index:
An index in which each record in the target database has a pointer to it.
Clustered indexes may or may not be dense. Nonclustered indexes are always dense.
dimension:
Those entities the users want to group by when exploring data in a data
warehouse, such as time, location, product type, etc. Dimensions are typically repre-
sented by tables in a star schema.
dimension block index:
A B+tree index (in DB2) used to access clustered dimension
blocks.
dimension table:
The smaller tables used in a data warehouse to denote the attributes
of a particular dimension such as time, location, customer characteristics, product
characteristics, etc.
disaster recovery:
Part of a DBMS (recovery manager) that provides log data of all
transactions executed since the last checkpoint or beginning of the current set of
active transactions. When a failure of some sort occurs, either hardware or software,
the log can be used to return to the same state of the committed transactions before
the failure occurred, thus allowing the committed transactions to survive. Typically,
uncommitted transactions are undone and must be restarted.
distributed data allocation:
Strategies concerning where to place fragments of data in a
system distributed over a computer network, including decisions about whether to
replicate the data and how much to replicate.
fact table:
The dominating table in a data warehouse and its star schema, containing
dimension attributes and data measures at the individual data level.
failover:
The ability for a secondary server to take over the processing responsibilities or
service requests that were previously directed to a primary server in the event the pri-
mary server goes offline. When a primary server fails and a secondary takes over, the
primary is said to have “failed over” to the secondary. Failover processing is a major
theme in high-availability computing.
