e in procuring data from other agencies, user must
purchase all the information compiled for that map
sheet and extract the feature relevant to the
applications, and
eo large volumes of datasets are currently not
maintained with any DBMS. They are maintained
simply as large sequential files on reels of tapes.
These and other related problems are what the federated
multi-database spatial information management system
(code named the Delta-X project) is concerned with. Such a
model of operation is not exclusive to GIS. Similar
problems have been addressed in purely corporate database
environments. The Delta-X is a multi-database system with
a common integrated global conceptual schema definition.
The global database is intended to achieve interoperability
between DBMS that have been specialized for three distinct
data types: spatial data (vector and raster), structured text
and free-text. Our design provides an integrated access to
data stored in relational databases, object oriented
databases, simple file systems and information retrieval and
document management systems. The current implementation
does this through Remote Procedure Calls by formulating
queries in the format of the target systems. Where the target
systems are relational DBMS, SQL is used as the language
for remote data access. Simple file systems are mapped into
relational tables. Thus, Delta-X establishes interoperability
between relational DBMSs and GISs.
To provide a user accessible data warehousing user
assistance for inventory and metadata review, a
complementary system: MetaView/GIS Spatial Browser
(MV/GIS) was developed as a front end to the Delta-X. The
MV/GIS and Delta-X systems are the basic component of
our Division's spatial data infrastrucure and provide a
solution to Geomatics Canada data warehousing.
2.. THE DELTA-X SYSTEM
2.1 The Delta-X Common Spatial Data Model
This is the conceptual data model into which other GIS
internal representations can be mapped. For vector data, the
Delta-X spatial data model maintains up to degree four
topology. ^ This global conceptual scheme forms the
intermediate transition schemes for data exchange between
different GISs. The global conceptual data model has a
mapping onto either a relational database, as a collection of
relational tables, or into an object-base as a colony of
categories. A Delta-X server maintains the vector data in a
relational database if the underlying database management
system is relational, e.g., Ingres, Oracle, Informix, etc.
Similarly, the Delta-X server retains the data in a set of
equivalent category classes in an object oriented database, if
the underlying DBMS is object-oriented, e.g., ODE,
ObjectStore, Objectivity. The significant idea is that Delta-
X server transparently delivers data to and receives data
from clients' GIS environment.
The illustration with the vector data shows the general
approach for handling data in Delta-X. This approach,
where by a common defined global data model is
materialized for actual representation in local databases
2
management system, is extended to handle the various
classes of data-type required in GIS. For example, raster
data such as remote sensed images, structured text that are
related to spatial objects, and free text are all represented in
the common global schema definition. Since some of these
data types have internationally defined standards of
representation, e.g., GIF, JPEG, MPEG, SGML, HTML,
etc., these are maintained as files in their respective
standard formats, and related to the defined spatial features
through spatial indexes. A number of such non-vector data
types are related to the spatial features using index
techniques.
2.2 Spatial Indexing
Even over a small area of coverage, maintaining all the
relevant information for all data types, quickly grows into a
significant large database. Delta-X organizes space into
hierarchical tessellated regions and maintains two distinct
levels of index schemes for identifying a feature in defined
space. The first level of index identifies a regular polygonal
cell of the region of coverage that has been hierarchical
tessellated into a near unicorn grid. The second level of
index associates features within each cell with other
information types. The first level uses a quadtree-like
method of spatial index. The second level draws from a
number of one-dimensional and multi-dimenesional index
techniques depending on the data-type to be indexed.
The partitioning of the space into cells (tiles), by the first
level of index enables the large volume of data to be
accessed and controlled in manageable units. By this means,
all data of a specific region or cell, and controlled by a
particular organization or agency, can be grouped together.
Geographic coordinate system (longitude and latitude) forms
the common and basic coordinate reference system by which
regions and features are related across multiple databases.
The local database may derive the actual projection and
coordinate reference system used in referencing spatial
objects within that particular environment.
2.3 The Delta-X Features and Services
Delta-X performs multiple client-server roles. First, Delta-
X servers control the data storage in commercial DBMS
systems. Second, it is a client-server transaction processing
system, and handles message and data exchange between
Delta-X database and/or other specialized servers and Delta-
X clients. Third, Delta-X performs specialized server
functions, such as data conversions to and from the internal
data interchange format. Finally, it acts as a proxy client on
behalf of GIS systems that produce or consume data - it is
the source or destination of data in Delta-X transactions
A single Delta-X server can be accessed by many Delta-X
clients at the same time. Even a single Delta-X client can
start many transaction on the same Delta-X server. In our
current implementation, all conversion processes share the
same CPU. The Delta-X server architecture, however, can
also be implemented on a parallel machine or on a cluster of
workstations on a dedicated LAN, where each transaction
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B2. Vienna 1996
pro
pro
Del
a cl
star
soft
Del
con
net
set
ope
lea!
net
on
moc
dat:
dat:
typ:
par
fed
eac|
use
spe
and
are
sen
not
Org:
ena
of t
whi
stor
Del
fun
mar
tran
reqi
tran
reqi
reqi
serv