Computer Science – Databases
Scientific paper
Oct 2007
adsabs.harvard.edu/cgi-bin/nph-data_query?bibcode=2007aspc..376....3k&link_type=abstract
Astronomical Data Analysis Software and Systems XVI ASP Conference Series, Vol. 376, proceedings of the conference held 15-18 Oc
Computer Science
Databases
1
Scientific paper
The Large Synoptic Survey Telescope (LSST), a proposed ground-based 8.4 m telescope with a 10 deg^2 field of view, will generate 15 TB of raw images every observing night. When calibration and processed data are added, the image archive, catalogs, and meta-data will grow 15 PB yr^{-1} on average. The LSST Data Management System (DMS) must capture, process, store, index, replicate, and provide open access to this data. Alerts must be triggered within 30 s of data acquisition. To do this in real-time at these data volumes will require advances in data management, database, and file system techniques.
This paper describes the design of the LSST DMS and emphasizes features for peta-scale data. The LSST DMS will employ a combination of distributed database and file systems, with schema, partitioning, and indexing oriented for parallel operations. Image files are stored in a distributed file system with references to, and meta-data from, each file stored in the databases. The schema design supports pipeline processing, rapid ingest, and efficient query. Vertical partitioning reduces disk input/output requirements, horizontal partitioning allows parallel data access using arrays of servers and disks. Indexing is extensive, utilizing both conventional RAM-resident indexes and column-narrow, row-deep tag tables/covering indices that are extracted from tables that contain many more attributes.
The DMS Data Access Framework is encapsulated in a middleware framework to provide a uniform service interface to all framework capabilities. This framework will provide the automated work-flow, replication, and data analysis capabilities necessary to make data processing and data quality analysis feasible at this scale.
Axelrod Tim
Becla Jacek
Cook Kelly
Gray James
Kantor Jeff
No associations
LandOfFree
Designing for Peta-Scale in the LSST Database does not yet have a rating. At this time, there are no reviews or comments for this scientific paper.
If you have personal experience with Designing for Peta-Scale in the LSST Database, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Designing for Peta-Scale in the LSST Database will most certainly appreciate the feedback.
Profile ID: LFWR-SCP-O-955610