Databases Reference
In-Depth Information
schema:
A conceptual data model that shows all the relationships among the data ele-
ments under consideration in a given context; the collection of table definitions in a
relational database.
secondary index:
An index structure that allows an efficient search to be made to a
table, given a Boolean search criterion, usually involving nonkey data (i.e., data val-
ues that are not unique).
sequential search method:
A simple search mechanism that searches every record in a
file or database.
shared-nothing partitioning:
Database systems based on relatively independent servers
that work cooperatively on subsets of a problem for high efficiency. When they need
to share data, it is passed between servers over a high-speed interconnect. They are
particularly useful in analyzing large complex data sets.
slice:
A block containing a single value of one dimension in a multidimensional data-
base.
sparse index:
An index in which each target data block or page has exactly one pointer
to it. Only clustered indexes can be sparse.
star schema:
The basic form of data organization for a data warehouse, consisting of a
single large fact table and many smaller dimension tables.
striping (or data striping):
A technique for distributing data that needs to be accessed
together across multiple disks to achieve a greater degree of parallelism and query
efficiency.
superkey:
A set of one or more attributes (data items) that, taken collectively, allow one
to identify uniquely an entity or a record in a relational table.
table:
In a relational database, the collection of rows (or records) of a single type (similar
to a file).
third normal form (3NF):
The most common form of a table considered to be suffi-
ciently normalized (broken up into smaller tables) to eliminate problems with
unwanted loss of data from the deletion of records (rows) and inefficiencies associ-
ated with multiple data updates due to redundant data. A table in 3NF has one or
more keys such that all nonkey attributes have exactly one value associated with each
key value.
unique index:
An index based on pairs of primary key values and pointers; each pointer
is the address for a block containing the unique record that has that specific key
value.
