Ibm’s Db2 Spatial Extender: Managing Geo-spatial Information within the Dbms

DATA TYPES. ADTs are user-defined types that describe complex column structures to the RDBMS. The new type has an encapsulated internal structure represented as a set of attributes. ADTs allow the user to represent new entities in the database, such as images, text, and spatial data. An example of a text ADT is a “document” data type with attributes such as text (a LOB that contains the actual text), language (character), and format (character). A sample spatial ADT is a “polygon” data type with attributes such as geometry (a LOB containing the points representing the polygon), area (integer), and number of points (integer). ESRI has defined specific spatial data types as ADTs in the Spatial Extender (see Figure 4). Users who need to define complex spatial types not yet supported by Spatial Extender, such as arcs, splines, ellipses, and circles, can define these as new ADTs or as sub-types of ESRI’s pre-defined ADTs. DB2 Spatial Extender Data Types Data Type Category Sample Applications Geometry Super class All geometries have the following attributes or properties: interior, boundary, exterior; simple or nonsimple; empty or not empty; number of points; envelope; dimension; Z coordinates; measures; spatial reference system Point Linestring Polygon Base geometry sub-classes Store or customer location Road, river, power line Water body, flood plain Multipoint Multilinestring Multipolygon Homogeneous collections sub-classes Incidents of epidemic outbreak Network of roads, earthquake fault line Non-contiguous parcel of land such as an island chain Figure 4 ADTs also support “value substitutability.” This is the ability to store different subtypes in the same column on a row-by-row basis. A column of type “geometry” can contain a point in one row and a polygon in another row if point and polygon are both sub-types of geometry. Without value substitutability, the data must be normalized so instances of each sub-type are stored in separate tables. This means, for example, that a query asking whether a point is contained within a particular polygon requires a join among three tables. In contrast, storing a variety of sub-types in one column avoids the necessity for joins when comparing spatial entities of different sub-types. This has two significant benefits. It enhances the developer’s flexibility in designing the database structure and improves performance. Because the contents of ADTs can vary dramatically in size, the DBMS can determine whether data should be stored inline in a column in the table, or stored ADTs use attributes to encapsulate internal structure Multiple sub-types can be stored in the same column Flexibility in storing ADTs IBM’s DB2 Spatial Extender 14 separately from the rest of the row in a manner similar to the way LOBs are handled. A point, for example, is stored inline as a single set of x,y coordinates. A polygon representing the state of California (many thousands of points) is stored separately. When defining an ADT, the user can specify a threshold, or number of bytes, that the DBMS stores inline. The DBMS relies on this threshold to move data between inline and separate storage on a row-by-row basis. For example, the user can choose to keep attributes inline for fast reference. If an ADT value is very small (e.g, a point), it will be stored inline. If the value is very large, it may be stored partially inline and partially in separate storage or all separately. TYPE HIERARCHIES. Another important feature of ADTs is support for subtypes and inheritance. By defining hierarchies of ADTs, the developer can reuse both attributes and methods. Figure 5 illustrates the type hierarchy in DB2 Spatial Extender and the fact that this hierarchy can be extended to incorporate new userdefined types, indicated by dotted lines. DB2 Spatial Extender Type Hierarchy