On Querying Data and Metadata in Multiversion Data Warehouse - Data Warehousing Design and Advanced Engineering Applications - page 210

Figure 1. An example DW version derivation tree consisting of real and alternative DW versions

DW Version Change Operations Every DW version is valid within a certain

time period represented by two timestamps, i.e.,
begin validity time (BVT) and end validity time
(EVT) (Bębel et al., 2004; Wrembel & Bębel,
2007), as shown in Figure 1. For example, real
version R 1 is valid from time t 1 (BVT) until t 2
(EVT) whereas R 4 is valid from t 7 until present.
Alternative versions are valid within the same
time period as their parent version, e.g., A 2.1 is
valid from t 3 until t 4 .

After being explicitly derived, a DW version may
be modified by means of 15 elementary operations
on a schema, further called schema change opera-
tions and by means of 7 elementary operations
on dimension instances, further called instance
change operations . These operations allow to
handle all typical changes that may be required
in practice. Multiple schema change operations
and instance change operations may be applied
to the same DW version.
Schema change operations include: creating a
new dimension, creating a new level, connecting
a level into a dimension hierarchy, disconnecting
a level from a dimension hierarchy, removing
a level, removing a dimension, creating a new
attribute for a level, removing an attribute from
a level, changing the domain of a level or a fact
attribute, creating a new fact, creating a new at-
tribute for a fact, removing an attribute from a
fact, creating an association between a fact and
a level, removing an association between a fact
and a level, removing a fact.
Instance change operations include: inserting
a new level instance, associating a level instance
with its parent level instance, associating a fact
instance with a bottom level instance, deleting

Metadata Describing MVDW

In order to provide the support for schema and data
versioning as well as for querying the MVDW , the
set of well defined metadata is required. In our
approach we developed the metaschema of the
MVDW (Wrembel & Bębel, 2007) that allows to
store data on: (1) the structure and content of every
DW version, (2) changes applied to DW versions,
(3) data conversion methods, which are required
for querying, (4) transactions run in the system.
The metaschema is general and it is applicable
to a ROLAP and a MOLAP implementation. It
is composed of dictionary tables .