The Horizon Cosmological Database

1
The Horizon Cosmological Database
http//horizon-vo.univ-lyon1.fr8080/horizon
2
The development team

• Romain Tessier CEA DAPNIA
• Bruno Guiderdoni CRAL
• Hervé Wosniak CRAL
• Julien Devriendt CRAL
• Jaime Forero CRAL
• Jean-Paul Le Fèvre CEA DAPNIA

3
The database schema

• Based on the Galics schema
• Normalized information is not duplicated,
  relationships are logically implemented
• Still not used yet on real data
• May cause performance problems

4
The Full Database Schema
Simulation
Boxes
Code params
Halo
Cosmology
Container
Ascendance
Descendance
Galaxy
Properties
Properties
Magnitudes
Cone
Apparent props
5
The Simulation Schema
Simulation
Cosmology
Box
cosmology_id
Simulation_id
timestep
author_id
User
parameters_id
Parameters
6
The Halo Schema
Halo properties
Halo _at_ ts
Ascendance
halo_id
properties_id
timestep
ascendant_id
container_id
simulation_id
Descendance
halo_id
descendant_id
7
The Galaxy Schema
Galaxy
comp properties
halo_id
galaxy_id
at_left
component (disc,bulge,burst)
at_rite
daughter_id
comp magnitudes
galaxy_id
filter_id
Filter
component
8
The database implementation

• MySQL
• But the application should stay db independant
• Connections managed by JDBC
• Servers located at Saclay, Cral, CC in2p3

9
The data ingestion

• File format is based on XML
• The content is specified by xml schemas
• Horizon-db.xsd, simulation.xsd, etc
• A file is parsed, checked and the content is made
  persistent in the database
• Running java hep.horizon.db.ie.SimulationLoader

10
Retrieving data from the database

• Some issues to address
• Volume of data is big
• Schema is complex
• Performances should be reasonable
• Security should stay sound

11
Accessing data through the web

• Examine what has been done by others
• Decide which kind of interface to implement
• Forms
• SQL
• Scripting language

12
SQL pros and cons

• Short learning curve
• Already known by many users
• Straighforward to write simple queries
• Hard to write complex queries
• Nightmare to debug
• Security is an issue

13
Bsh pros and cons

• Reasonable learning curve like java, python
• These languages are very popular
• True programming language
• A huge number of APIs available
• Its a new language still evolving
• The best scripting language is not known yet

14
Bsh how it looks like

• haloNumber 2
• simulationName "test"
• handler new SimulationHandler()
• halo handler.selectHalo(simulationName,
  haloNumber)
• list halo.getSortedGalaxies()
• for (Galaxy g list)
• print(g.getNumber())

15
Output results

• If only a few lines output on the web page
• Otherwise in a csv formatted ascii file which can
  be downloaded by http
• Fits files should be available in a near future

16
The Horizon Cosmological Database
internet
MySQL
Tomcat
Client
Application Server
Database Server
Database
browser
http
Client
jdbc
data
Java programs
cayenne
Mozilla
Client
IExplorer
17
Its a work in progress

• A lot of things still left to be done
• At this point we are testing the prototype
• Access library should be developped
• SQL BSH scripts should be written and run
• First simulation results should be loaded
• GUI, documentation should be improved
• User management should be implemented
• Super big simulations should be loaded