ALMA Science Data Model

ALMA Science Data Model
(ASDM)

History:

The ASDM inherits from the data model of the MeasurementSet.

AEDF renamed ASDM:
1. To highlight that this model is intended for all the ALMA data which are useful specificaly in a scientific context
2. To avoid confusion with the TBD export (i.e.outside the ALMA domain) format
3. AEDF version 1 (initial draft) had been examined by M. Wright and L. Testi and discussed at the last SSR face to face meeting
4. AEDF version 2 (official document):
  - preserves the concept of vers. 1
  - has a significant modification being restructured at some level (reasons explained in a document on the Twiki)

Defining the ASDM:

A process in 3 steps:

Where we are?

Conceptual schema mostly done:

refinements will be needed in various places

iterative process while implementing and using the model
- tests via simulations
- tests to host real data from existing millimeter instruments
"MEASURE" column needs to be described
e.g. ACSTime, ...

some properties needs to be described explicitly

a number of "values" need to be reserved

Definition of the logical schema:

The conceptual schema may be designed in different forms

The model is presented using the form of a set of relational tables.

But this does not imply that the logical model must be implemented using the classical E/R schema approach.

There are several different possible paths and technologies for the implementation.

A prototype implementation has been developped by the offline subsystem.

objective: fast implementation

to experiment quickly with the model

to experiment the concept of a Data Capturer
For R2 DC/Control does not yet use all features, e.g. more than one subarray in an ExecBlock

to have archived simulated data which can be used by other subsystems (require also the development of a filler)

to experiment with queries to investigate options for the actual archiving

an "original" path: the root of the automated code generation is the AEDF specification document itself:

used from a simulator to archive full datasets in the XMLStore

used for the DC (R2)

used by several subsystems: Control, Correlator, QuickLook, TelCal

The HLA implemention

Path: automated code generation and documentation using as the root an UML class diagram.

Work started in July 2004; work in progress

From the logical to the physical model

Requirements for Queries

Denormalization

The ASDM:

Facts:

The ASDM appears as very complex for some people!

This appearance is due to the large number of relations between tables
Tables being flat 2D, keys are composed of identifiers to provide the depth.

As a number of tables are normalized, there is in the ASDM multi-dimensional structures.

Seen as such the complexity is essentially removed.

The ASDM is actually composed of
1. multi-dimensional structures (for the meta and auxiliary data)
2. the configDescription:
  - a versatil connector
  - set the shape of the data cell
3. a tree (i.e. hierarchical) structure (the project structure)
Incremental view of the ASDM

The ASDM dataset:

The archived dataset corresponds to an ExecuteBlock

The dataset in the Science DataBase

The ASDM datset contains two external references: the project identification and the data_oid

The ALMA Science Archive requires more links in order to concatenate datasets into a dataset. These links are known as the ASDM is also used in the Telescope domain.
(see discussion on the Calibration Data Base)

Why the OT ignores the ASDM?
- The OT does not support all ASDM capabilities (see Target!
- Would make the life easier for CONTROL
- Would be convenient for the ALMA Science Archive (see SchedBlock)!

Export Data Format:

Requires following the activity on the VO side!