Keywords: DataAccWG
Hi,
After quite a few long discussions with Kem, I think we are starting
to get a picture what changes are needed in the prototype schema
(llnl+uw merged schema) to reflect:
1) assumptions made in the database_size_estimate spreadsheet
2) proposed optimizations
3) fixes to the recently uncovered problems.
Here is a summary:
Object table
============
contains information about stars (multi-epoc)
Changes needed:
- add index on (ra,decl)
- add 5 columns (1 for color precomputed as
u-g, g-r, r-i, i-z, z-y). Color column type:
decimal(5,3), so that is 15 new bytes
- need to decide if we need to keep magnitudes
or assume they will be reconstructed from colors
- I'm currently assuming we keep magnitudes,
and add 5 colors, so that is 192 bytes/row
- I wonder if we should keep that name or rename it
to "Star" or "Stars"
Galaxy table
============
one entry (a "source") per galaxy containing average
value of all corresponding detections, need to store
information for all 5 colors.
This table does not exist in the current schema.
Estimated size: 878 bytes/row
MovingObject table
==================
we have that in schema, probably ok as is
VarObj table
============
Contains variable objects: ~5% of number of rows in Object table.
This table does not exist in current schema.
Estimated size: 100 bytes/row
Source table
============
Contains all (non-DIA) detections.
At the moment in our scheme Source table has galaxy related info
(it incorrectly assumes that galaxies will be processed for every
frame, the photometry group decided to do that only on
stacked frames for galaxies, and not per frame). So the existing
Source table is a good foundation for Galaxy table, except that
we need to add info for each color.
We need to create a separate table for sources related to stars.
Estimated size: 70 bytes. We will probably call it "Source"
deep db / temp db
=================
So if we make the changes outlined above, then:
- "deep db" will correspond to tables: Star, Galaxy, VarObj
- "time db" will correspond to tables: Source
Other changes
=============
- need to add postage stamp support
- need to add template images support
Related links
=============
Latest version of merged (LLNL+UW) prototype schema (temporarily
in sandbox until I find a better place of it)
https://www.lsstcorp.org/docushare/dsweb/Get/Document-1888/mergedSchema.sql
DB storage estimates:
https://www.lsstcorp.org/docushare/dsweb/Get/Document-1779/lsst_storage_estimates.xls
I am going to assume the proposed schema layout for further disk io
analysis.
Any questions or comments?
Jacek
_______________________________________________
LSST-data mailing list
[email protected]
http://www.lsstmail.org/mailman/listinfo/lsst-data
