Databases Reference
In-Depth Information
OLAP hierarchy:
For each dimension defined for data in an OLAP system, the
attributes of each dimension can be structured as a hierarchy. For example, time can
be broken down into years, months, weeks, days, etc. Each level in the hierarchy can
be used to group data into materialized views.
OLTP:
Online transaction processing; the preferred method for database processing
before data warehouses.
partitioning:
An algorithm for reducing the workload on any one hardware component
(like an individual disk) by dividing the data fairly evenly over several components.
This has the effect of balancing the workload across the system and minimizing bot-
tlenecks.
physical block:
See block.
physical database design:
The step in the database life cycle involved with the physical
structure of the data, that is, how it will be stored, retrieved, and updated efficiently.
In particular, it is concerned with issues of table indexing, partitioning, and cluster-
ing on secondary storage devices (e.g., disks).
primary key:
A key that is selected from among the candidate keys for an SQL table to
be used to create an index for that table.
query execution plan (QEP) (or query plan):
An extended relational algebra tree
specifying the set of precise relational algebra operations (in a specific order) needed
to complete a query.
query execution plan selection:
In query optimization, alternative query execution
plans are evaluated for estimated cost (usually in I/O time) and the plan with the
minimum estimated cost is selected for execution.
query optimizer:
Software within a DBMS that determines the fastest way to execute a
query through the analysis of query execution plans and that estimates the I/O time
needed to execute each candidate plan.
query response time:
The real time between the last keystroke of a query and the last
character displayed in the response; it is typically composed of the I/O time, CPU
time, and network time, including delays, and often overlapped.
RAID:
Reliable array of inexpensive disks. Replacement of expensive, high-performance
devices with arrays of lower cost disks to achieve similar high performance and reli-
ability by the use of parallelism, striping, redundancy, and parity.
range partitioning:
The partitioning of a table by selecting ranges of attribute values
such that the table can be divided into nearly equal numbers of records. Each range
of records is stored on a separate disk to help balance workloads and maximize paral-
lelism.
