LSC Data Analysis Software Working Groups

Project Description

The LIGO Data Analysis System - (LDAS) is the LIGO Laboratory's integrated software/hardware data analysis system for analyzing and processing of LIGO Frame data downstream of the generation of the archival Frame files. The system is based on a collection of distributed processes that are centrally controlled to carry out complex tasks across multiple compute servers by flow control known as a pipeline job. The distributed nature of the processes involved in the pipeline allows for concurrency ( typically on the order of 30 pipelines) within a given LDAS system. The flexibility of the distributed processes allows for many distinct flavors of pipeline jobs to be carried out within LDAS. Here is just a few examples of the classes of pipeline jobs available (details here):

• Get raw frame data products
• Query the internal LDAS database (data-mining)
• Insert metadata (triggers, search parameters, etc.) into the database
• Perform complex signal processing on raw data and capture results
• Perform complex detector characterization studies
• Carry out complicated astrophysical searches and store results to the database or to files
• Generate Reduced Data Sets (RDS Frames) from Frame data

The LIGO Laboratory operated nearly a dozen LDAS systems. The principle systems available for use by the LIGO Scientific Collaboration are located at Caltech, MIT and the two observatories. These system can be monitored by simply pointing a web browser at the home page's of the LDAS systems and following the links to the current jobs, API logs, statistics and other monitoring pages. The links for these four primary LDAS web sites are:

• California Institute of Technology (CIT)
• Massachusetts Institute of Technology (MIT)
• LIGO Hanford Observatory (LHO)
• LIGO Livingston Observatory (LLO)

In addition, the LDAS development team operates seven other LDAS systems at Caltech for development and testing purposes. The two most closely matching the hardware configurations of the systems above are:

• (LDAS-DEV)
• (LDAS-TEST)

NOTE: It is possible to move around from one LDAS system to another simply by clicking on the map of the lower 48 States located at the top of any LDAS home page.

Project Contacts

  Kent Blackburn (blackburn_k at ligo dot caltech dot edu): Architect
  Stuart Anderson (anderson at ligo dot caltech dot edu): System Administration
  Ed Maros (emaros at ligo dot caltech dot edu): Librarian

Downloads

Official Releases

The latest official release of LDAS is version 1.7.0, released on July 27th, 2005 by LIGO's LDAS development team at the California Institute of Technology. A list of changes since the version 1.6.0 release can be found at the following website. Downloads of this release are available in two formats:

• Binary distribution for SunOS 5.9 & Fedora Core 3 Linux via mirroring
  
• Tarball distribution of source code download from LDAS website
  

In order to run and operate an LDAS system, you will also need the LDCG tool suite and LAL/LALwrapper. These packages can be mirrored along with the binary distribution of LDAS over the internet, or if one prefers, it can be built using the official "how-to" document found at the LDAS website.

To fully utilize a running LDAS system requires the installation of IBM's DB2 database server. This database is available under an academic licensing agreement for teaching and research known as the IBM Sholars Program at no charge. Once you have acquired the necessary licensing from IBM, you will find instructions for installing and using the DB2 database on the LDAS website.

LDAS is a large complex software system designed to run on a large distributed network of servers and PC clusters. However, due to the shared object technologies used by LDAS processes, it is quite possible to run LDAS on single computer whether it be a desktop or even a laptop. In fact, the performance of LDAS runing on a single computer is quite remarkable when the types of requests and the rate at which requests are being made does not place a huge demand on the single computer's memory resources. A custom "how-to" documenting the procedure for setting up LDAS on a single computer is available at the LDAS websites for those interested in running a mini-LDAS at home. It is recommended that this be installed on a single computer with a clock speed of at least 2GHz and at least 1GB of memory for best results.

Sub-Packages

There are a number of subpackages that contain complete self contained subsets of the full LDAS software distribution (Excluding needed LDCG Software!).

• LDAS This is the complete LDAS software tarball.
• cmonClient This is the controlMonitorAPI's GUI tarball.
• frameCPP This is the C++ frame I/O library tarball.
• wrapperAPI This is the standalone wrapperAPI tarball.
• External DataCondition Library This is a library that can be used external to LDAS and provides the same signal processing algorithms available within LDAS's dataConditionAPI.
• External DiskCache Library This is a library that can be used external to LDAS and provides the frame data discovery engine for caching gravitational wave datasets stored in frames on any mounted file system. This particular subpackage is also available in binary formats for Fedora Core 3 Linux and Sun Solaris 9.

These are only available as tarball distributions of the source code. Documentation for how to get and install these subpackages, along with their dependencies on LDCG tools, can be found at the following website.

Development Snapshots

The LDAS development system website posts nightly builds of tarball distributions of the development code base that is current in CVS as of 7:00 PM Pacific Time. The snapshots remain available for up to ten days before they are deleted in favor of newer snapshots. There is no guarantee that any of these development snapshots are public use, but they are made available for those interested in participating in LDAS development or have a need for functionality on present in the devlopment code. No support for these snapshots is intended.

The development snapshots of LDAS require the development version of the LDCG tools. This version of LDCG is experimental and not available for mirroring over the internet. Instructions for how to build the development version of LDCG can be found in the LDAS development "how-to" document.

Bug Reporting, Testing, and Quality Assurance

Problem Tracking

Access to the LDAS GNATS web based problem tracking system is password protected, requiring a username and a password to gain access. To apply for an LDAS GNATS website username and password, please fill in the requested information at this URL.

Once you have the username and password you can access the Problem Tracking System at this URL. Be sure to select the "ldas" Database from the pulldown menu at the website. In addition, if you plan to file a Bug Report, please review these general guidelines for Severity, Priority and Class and also review the current open problem reports to verify that your bug hasn't already been filed. If your bug exists and has not been fixed, add additional information to the existing bug if available. If your bug exists and has been fixed, upgrade to the version in the bug report to determine if the bug was properly fixed. If it was not, reopen the bug.

Unit, Integration & System Testing

The LDAS Software Development Team routinely carries out unit(component level), integration(program interface level) and system(user interface level) testing of the LDAS Software System. Results from these tests are available on the web at the following links:

• Nightly Unit Testing History of unit tests performed each night on current CVS snapshot.
• Current Integration & System Testing Most recently performed integration and system testing.
• Historic Integration & System Testing Performed routining and as part of the release preparations.

Using the controlMonitorAPI, it is also possible to perform many of the system level tests in a one-shot mode from the client's GUI. Under the LDAS Test menu are sample user commands to exercise each user command. It is also possible for the experienced user to make edits to the existing scripts to carryout user defined commands. Results of these system tests are reported back to the controlMonitorAPI's client once the test job completes. For more information, see the Using controlMonitorAPI Client webpage found off of any LDAS homepage.

Quality Assurance

The LIGO Development Team has conducted a series of mock data challenges in collaboration with the LIGO Scientific Collaboration. The details of these MDCs are captured in technical documents that can be found in the LIGO Document Control Center:

• LDAS Data Conditioning API: Mock data Challenge
• MPI Mock Data Challenge
• Summary of the LDAS Database Mock Data Challenge
• Inspiral Mock Data Challenge (Not available in LIGO Document Control Center)
• Known Pulsar Demodulation Mock Data Challenge
• Burst/Stochastic Mock Data Challenge (Electronic version not available in LIGO DCC)

These MDCs provided many of the integration and system level test scripts that are used in the LDAS test procedures. As a result of these efforts the LDAS Development Team was able to work closely with the LIGO Scientific Collaboration to assure that the software and hardware platforms of LDAS meet the quality assurance needs of the community.

Mailing Lists

• LDAS Software Support
• LDAS System Administration Support

Documentation

• Homepage
• Instructions for applying for an LDAS account
• LDAS User Command Instructions
• Guidelines for installing LAL and LALWrapper for use in LDAS
• LAL-LDAS Interface Coding Specification
• LALWrapper software development document
• How to develop search code to run under the LDAS wrapperAPI

Open Tasks

Next Relese

Here is a list of tasks that are scheduled to go into the next release of LDAS.

Long Term

• Port LDAS and LDCG C/C++ source code to Solaris C/C++ compilers ( Forte).
• Port LDAS and LDCG C/C++ source code to Intel C/C++ compilers
• Implement federated/replicated database WAN between LDAS systems.
• Develop a dataConditionAPI shared object for use in environments other than LDAS. (available in LDAS 1.1.0 - 1.7.0 releases)
• Develop a diskCacheAPI shared object for use in environments other than LDAS. (available in LDAS 1.2.0 - 1.7.0 releases)
• Extend actions supported in dataConditionAPI's algorithm{} grammer.
• Provide branching syntax (if-then-else).
• Provide looping syntax (foreach & for(i:1:N) ).
• Port to true 64 bit LDAS binaries.
• Develop multi-frameAPI topologies.
• Replace outdated C++ socket communications library with open source libraries (ACE).
• Provide data product visualization through controlMonitorAPI GUI.
• Currently exists for data products inserted into database.
• Extend to support LIGO_LW and Frame formats.
• Work with TclGlobus Project to provide globus support in LDAS.
• Provide user authentication/authorization through certificates.
• Provide data product publication onto the GRID.
• Implement strong data validation support in frameAPI.
• Checksum comparision in datapipelines. (available in LDAS 1.2.0 - 1.7.0 for createRDS only)
• Test dataValid flag in FrADC channels. (available in LDAS 1.2.0 - 1.7.0 for create RDS only)
• Timestamp validation between filename and channel content. (available in LDAS 1.2.0 - 1.7.0 for createRDS only)
• Closed out problem reports and enhancement requests (ongoing).
• Maintain compatability with currently supported software tools (ongoing).
• Enhance online documentation (ongoing).

Contributors (short list)

Anderson, Stuart
Barnes, Masha
Blackburn, Kent
Charlton, Philip
Ehrens, Phil
Finn, Sam
Kilmenko, Sergei
Lazzarini, Albert
Lei, Mary
Maros, Ed
Mendell, Greg
Romano, Joe
Salzman, Isaac
Searle, Antony
Sears, Bruce
Shawhan, Peter
Turner, Mark
Yakushin, Igor
Zweizig, John

Last Modified Wednesday, August 3rd 2005 by Kent Blackburn

$Id: ldas.html,v 1.74 2006/01/25 22:10:28 emaros Exp $