Transcript Building & Extending Applications with DataBlade Modules

An HDF-EOS DataBlade
using Informix’s ObjectRelational Database
Renu Chaudhry
ECOlogic
www.ecologic.net
Topics
 What is an ORDBMS ?
 INFORMIX-Universal Server Architecture
 DataBlade Architecture
 HDFEOS DataBlade Components
 HDFEOS DataBlade Benefits
 User Scenario
 Summary
What is an ORDBMS ?
ORDBMS : Object-Relational Database Management System
 OO features
complex data object
data behaviors (functions)
inheritance and overloading
 RDBMS features
flexibility, security, transactions, recovery
Native SQL parsing, optimization
data access methods (use of R-tree & B-tree indexes)
INFORMIX-Universal
Server Architecture
Image
Spatial
Web
Extending the server by
adding DataBlades
DataBlade API
Text
Extensible
OR Engine
Scaleable
Data
Manager
INFORMIX-Universal Server
 Data Types to define new information
or data structures
Types
 Routines that act upon the new data
types
Routines
 Tables containing new data types
 Fast indexing and access for new
data types
 Client API
Tables
Access
Methods
Client
Code
DataBlade Module
DataBlade Architecture
HDFEOS Datablade
New
Types
data types: Grid, Swath, Point
Routines:
data inquiry functions,
subsetting functions as referred in HDFEOS library
Routines
HDFEOS
Tables
indexing methods for sorting
and storing data
Tables
containing HDF-EOS metadata
and data
Client
Access
Methods
oriented code for data processing
Client
Code
DataBlade Module
HDF-EOS Datablade components
will include:
HDFEOS DataBlade
Benefits
 Access cross granule information via SQL
 Easy browsing
 Improved performance with use of built-in index
 Portability - data and functions packaged inside database
server
 Ease of integration with other data types
INFORMIX-Universal
Server
Queries
Results
HDFEOS User Scenario
A user wishes to perform a correlation study between different physical
parameters for a specific geographical area.
example: Correlation between cloud amount and precipitation
 User issues query to view all grid data objects that covers Bermuda.
 The query returns all granule names, grid information (names,
dimensions, projection, fields etc.) matching the criteria.
 User issues query to retrieve data of interest.
 Query returns data of interest
Sample Scenario
select Grd.filename, Grd.name, Grd.projection, Grd.dimensions, Fld.fieldname
from HEOSgrid Grd, HEOSgdfield Fld
where ((Grd.filename = Fld.filename and Grd.objid = Fld.gridid) and
Contains(Grd, '(32, 70, 42, 60)');
User Scenario cont’d
 Query to retrieve subset of data (“Precip” field)
select LatLonSubset(Fld, '(32, 70, 42, 60)')
from HEOSgdfield Fld
where Fld.filename = 'pathfinder_flat_month.8708.psg.hdfeos' and
Fld.gridname = ’SSMI'
and
Fld.fieldname = ’Precip';
 Query to retrieve entire granule
 execute HEOSretrv((granulename, outfile , location);
HDFEOS Datablade
Summary
 Easy access and manipulation of HDF-EOS grid, swath, point data
types via SQL3
 Data and the functions reside inside database for better
performance
 Structural metadata search across granules by simple SQL
 Only SQL knowledge required by novice USERS to access HDFEOS data
 Database provides data reliability, security, portability, extensibility,
transaction, performance
 Datablade provides flexibility to incorporate HDF data objects
HDFEOS User Scenario
A user wishes to perform a correlation study between different physical
parameters for a specific geographical area.
example: Correlation between cloud amount and precipitation
 User issues query to view all grid data objects that covers Bermuda.
 The query returns all granule names, grid information (names,
dimensions, projection, etc.) matching the criteria.
 User issues query to view fieldnames from granules/grids of interest
 Query returns fieldnames contained in the grid
 User issues query to retrieve data of interest.
 Query returns data of interest
Sample Scenario
select * from HEOSgrid Grd where Contains(Grd, '(32, 70, 42, 60)');
Scenario - select fields
 Query to list all fieldnames contained in the specified grid objects
select filename, gridname, fieldname
from HEOSgdfield
where ((gridname = ‘SSMI’ and filename = ‘xxxx’) and
(gridname = ‘CloudAmt’ and filename = ‘yyyy’));
OR
select filename gridname fieldname
HDF-EOS search queries
 To get summary of all filenames, their size and the number of
objects in each data file residing in the database:
 select orgfile, size, numobj from HEOSfile;
To retrieve core, product specific, or archive metadata from
specific HDF-EOS data file:
 select core-metadata, product-metadata from hdfeos where
orgfile=‘toms_monthly_498.HEOSfile’;