MIDDLEWARE
Developing Data Solutions at NCSA
By Trish Barker, NCSA -- Sensors and instruments are the foot soldiers of science. They gather and generate the data that fuels investigation of phenomena ranging from the chemical reactions in rippling coastal waters to the energy rippling from quasars in deep space. 
While the data provided by sensors and instruments are a boon, managing, processing, and storing the flood of data presents a challenge. NCSA is collaborating with the National Optical Astronomy Observatory (NOAO) to develop solutions for managing the tens to hundreds of gigabytes of data generated each night by its observatories. Using these solutions as first steps, the ultimate goal is to meet the needs of the Large Synoptic Survey Telescope (LSST); when LSST begins operation in 2013 it will generate an estimated 15 terabytes of raw data and more than 100 terabytes of processed data every night, 365 days a year. NCSA leads two of the three LSST data management teams and is responsible for managing the integrated work plans, tracking progress, reporting to the LSST Corporation, and coordinating a series of data challenges. "Our NOAO/NCSA relationship provides very natural growth to LSST," says Chris Smith, astronomer and manager of the Data Products Program at NOAO, the group in charge of developing and operating the NOAO data management system. "There are large datasets accumulating today, and really cutting-edge developments are being implemented and are benefiting scientists tomorrow rather than in 2013." Working closely with domain scientists is a hallmark of NCSA's approach to the development of cyberenvironments -- which encompass and integrate distributed computing and data resources, scientific application codes, workflow tools, and user-friendly interfaces into end-to-end scientific processes. NOAO brings scientific depth to the collaboration, while NCSA brings both technological expertise and experience in supporting astronomy research. "We can talk to each other very well, and we complement each other very well," Smith says. "The great thing about this partnership with NOAO is we have a living, breathing community to work with," says Ray Plante, a senior research scientist at NCSA and the leader of the center's LSST effort. A larger telescope, and more data The Large Synoptic Survey Telescope project can be broken into three parts, each presenting challenges: the telescope, the camera, and the data management system. The 3.2-gigapixel camera will be the largest ever built. The telescope itself must be both very large (with an 8-meter mirror) and provide a very wide field of view; most mirrors of that size are designed to have a small field of view. "It's a novel telescope design," Smith says. The telescope's three heavy mirrors will need to be stringently aligned, "and of course you can't just do it in a fixed configuration," Smith adds. "You have to move it and you have to move it fast" in order to survey the sky. "It's challenging enough that they can't even develop the final design until they know what mountain it will go on, because the foundation under the telescope matters," he says. When the camera and telescope are completed and operational, LSST will image the entire viewable sky every three days. Its comprehensive, time-lapse imaging will provide an unprecedented census of the solar system, including transient objects like comets and potentially hazardous near-Earth asteroids. That prospect excites Smith, whose own research deals with supernovae, and other astronomers. "This system will be a great driver for the discovery of unexpected things," he says. "There's a lot of needles in the haystack that we haven't been able to find because the haystack isn't big enough." More detailed observations expand the haystack of data that astronomers are searching. LSST's repeated sweeps of the sky will also help to reduce noise, allowing astronomers to home in on fainter and fainter objects; by seeing farther and farther into the universe they are also seeing further and further into the past. "This is going to be very important in understanding the early universe," Plante says. And in perhaps its most mysterious challenge, LSST aims to provide crucial clues about the nature of "dark energy," the enigma that is causing the expansion of the universe to accelerate. In order to gain these insights, however, scientists must manage and analyze the incoming data. The 15 terabytes of raw images generated each night must be transmitted from the telescope and processed—generating a combined total of raw and processed data products that will top 100 terabytes each night. The data pipeline must also allow for quick, near real-time processing in order to provide feedback to the telescope to optimize imaging and to promptly alert the astronomy community about interesting observations, allowing researchers around the world to turn their telescopes toward the detected phenomena at the right time. NCSA provides data archive To lay the foundation for the LSST cyberenvironment, NCSA and NOAO are developing a prototype data pipeline using the vast stores of data generated by the ground-based observatories NOAO oversees: Kitt Peak in Arizona, Cerro Tololo Inter-American in Chile, and the Gemini Science Center, with observatories in Chile and Hawaii. NCSA created a mirror of the NOAO data archive in Urbana. The archive replication system is built on the Storage Resource Broker (SRB) middleware developed at the San Diego Supercomputer Center (SDSC) and a transfer queuing system NCSA developed for its archive of data from the Berkeley-Illinois-Maryland Array (BIMA) radio telescope. Mirroring NOAO's archive not only gives astronomers a high-bandwidth site from which to access data, it is also part of a strategy of security through redundancy, ensuring that data will survive at one site even if a catastrophic event hits another site. NCSA's robust mass storage system archives more than 2.3 petabytes of researchers' data, with data added at a rate of 40 to 60 terabytes each month. Michelle Butler, the manager of NCSA's Storage Enabling Technologies group, says the data management expertise of the center's staff-encompassing experience with running large file systems, parallel file systems, many storage architecture types, HPC storage, and database storage -- is as much of an asset as the storage infrastructure. "NCSA storage is designed with a long-term view," Plante says. "I can go back to data that's nearly as old as NCSA and have confidence that it will be there." Archiving the data at NCSA also will enable astronomers to take advantage of the center's 41 teraflops of high-performance computing power to process the data. Raw observatory data is marred by artifacts of the observing site, the weather conditions, and the instrument itself; it's as though each image is blurred by fingerprints that need to be wiped away so astronomers can get a sharp view of the data. "We want to get a picture as if it were taken in space, so we have to take out the effects of the telescope and take out the effects of the atmosphere," explains Plante. Tools for accessing data As challenging as the tasks of capturing, moving, processing, and storing data are, they are just the preliminary steps. The real excitement begins when astronomers can access and analyze the data. For LSST, community access will be provided through a Web-based virtual observatory (VO). Therefore, NOAO and NCSA are working to develop the VO model and VO tools, including an authentication and authorization framework for the NOAO portal, an online tool that enables users to find, access, and analyze the data available through multiple public archives, such as the Sloan Digital Sky Survey, Canadian Network for Observational Cosmology, Chandra X-ray Observatory Center, and others. "The NCSA/NOAO collaborative effort provides a backbone for secure data access, which is a vital component for astronomical portals and multi-location image archives," says Chris Miller, an assistant astronomer at Cerro Tololo Inter-American Observatory and one of the collaborators on the NOAO portal project. "These security measures are currently missing from ... most astronomical archive tools and services." By leveraging grid technologies -- including the Globus Toolkit, NCSA's MyProxy, and PURSe (Portal-based User Registration System) -- NCSA and developers from both Argonne National Laboratory and the National Virtual Observatory project have simplified the authentication and authorization process for users. When users log in, the portal knows which services and data they can access and which are withheld from them. This prototype portal was demonstrated in January at the American Astronomical Society Meeting in Washington, D.C. "Our demo included the ability for users to register and log on so that only they could see their proprietary data," Miller explains. "The next version, scheduled to be complete by July, will have a fully implemented security model developed by NCSA in collaboration with the NOAO." The grid-based security framework will serve as an international standard for virtual observatory interoperability. "The development of this security model is fully within the standards and procedures defined by the International Virtual Observatory Alliance," Miller says. "Thus, led by Ray Plante and Ramon Williamson at NCSA, and in collaboration with the NOAO Data Products Program Team based in Chile, we are leading the nation in applying a standard user authentication and authorization model to astronomical portals, archives, and analysis services." Testing the prototype pipeline In July 2006, the LSST project began its first Data Challenge, a test designed to evaluate the development of the data pipeline. The annual challenges will provide feedback for the team and will help the collaborators further refine the requirements of the LSST pipeline. When the observatory begins operation in 2013, data will move from the telescope to a nearby base camp, where some limited data processing will take place in order to provide feedback to the telescope and rapid alerts to the astronomy community. From the base camp, raw data will be transmitted to archive centers for processing, storage, and dissemination to astronomy researchers. Three sites in the National Science Foundation-funded TeraGrid network are standing in for the telescope (Texas Advanced Computing Center), the base camp (San Diego Supercomputer Center), and the archiving center (NCSA). Data will be transferred from site to site and processed along the way in order to evaluate the design of the prototype data management system. This prototype integrates grid technologies with components developed by NCSA's partners at the LSST Corporation, the National Optical Astronomy Observatory (NOAO), the Stanford Linear Accelerator Center, and the University of Washington. First, the challenge will test the data replication software, which is used to transfer data from site to site. Developed at NOAO, the Data Service (DS) software leverages SDSC's Storage Resource Broker (SRB) software. Then the basic functionality proposed for the data processing pipeline will be evaluated using prototype science codes and "resource consumers" that model how actual algorithms would consume compute cycles. These codes will be stitched together through middleware components developed by NCSA and its partners, to mimic the actions and applications that will be components of the final pipeline. "The challenge mimics the data transport and processing as it will happen in real life once the telescope is operating," says Cristina Beldica, project manager for NCSA's LSST effort. "What we're striving to do is get a big enough volume of data to show that our system will scale to the unprecedented data rates we'll see when the telescope is operational." This research is supported by the National Science Foundation. The effort to build the LSST is overseen by the LSST Corporation. For more information, visit its Web site.
While the data provided by sensors and instruments are a boon, managing, processing, and storing the flood of data presents a challenge. NCSA is collaborating with the National Optical Astronomy Observatory (NOAO) to develop solutions for managing the tens to hundreds of gigabytes of data generated each night by its observatories. Using these solutions as first steps, the ultimate goal is to meet the needs of the Large Synoptic Survey Telescope (LSST); when LSST begins operation in 2013 it will generate an estimated 15 terabytes of raw data and more than 100 terabytes of processed data every night, 365 days a year. NCSA leads two of the three LSST data management teams and is responsible for managing the integrated work plans, tracking progress, reporting to the LSST Corporation, and coordinating a series of data challenges. "Our NOAO/NCSA relationship provides very natural growth to LSST," says Chris Smith, astronomer and manager of the Data Products Program at NOAO, the group in charge of developing and operating the NOAO data management system. "There are large datasets accumulating today, and really cutting-edge developments are being implemented and are benefiting scientists tomorrow rather than in 2013." Working closely with domain scientists is a hallmark of NCSA's approach to the development of cyberenvironments -- which encompass and integrate distributed computing and data resources, scientific application codes, workflow tools, and user-friendly interfaces into end-to-end scientific processes. NOAO brings scientific depth to the collaboration, while NCSA brings both technological expertise and experience in supporting astronomy research. "We can talk to each other very well, and we complement each other very well," Smith says. "The great thing about this partnership with NOAO is we have a living, breathing community to work with," says Ray Plante, a senior research scientist at NCSA and the leader of the center's LSST effort. A larger telescope, and more data The Large Synoptic Survey Telescope project can be broken into three parts, each presenting challenges: the telescope, the camera, and the data management system. The 3.2-gigapixel camera will be the largest ever built. The telescope itself must be both very large (with an 8-meter mirror) and provide a very wide field of view; most mirrors of that size are designed to have a small field of view. "It's a novel telescope design," Smith says. The telescope's three heavy mirrors will need to be stringently aligned, "and of course you can't just do it in a fixed configuration," Smith adds. "You have to move it and you have to move it fast" in order to survey the sky. "It's challenging enough that they can't even develop the final design until they know what mountain it will go on, because the foundation under the telescope matters," he says. When the camera and telescope are completed and operational, LSST will image the entire viewable sky every three days. Its comprehensive, time-lapse imaging will provide an unprecedented census of the solar system, including transient objects like comets and potentially hazardous near-Earth asteroids. That prospect excites Smith, whose own research deals with supernovae, and other astronomers. "This system will be a great driver for the discovery of unexpected things," he says. "There's a lot of needles in the haystack that we haven't been able to find because the haystack isn't big enough." More detailed observations expand the haystack of data that astronomers are searching. LSST's repeated sweeps of the sky will also help to reduce noise, allowing astronomers to home in on fainter and fainter objects; by seeing farther and farther into the universe they are also seeing further and further into the past. "This is going to be very important in understanding the early universe," Plante says. And in perhaps its most mysterious challenge, LSST aims to provide crucial clues about the nature of "dark energy," the enigma that is causing the expansion of the universe to accelerate. In order to gain these insights, however, scientists must manage and analyze the incoming data. The 15 terabytes of raw images generated each night must be transmitted from the telescope and processed—generating a combined total of raw and processed data products that will top 100 terabytes each night. The data pipeline must also allow for quick, near real-time processing in order to provide feedback to the telescope to optimize imaging and to promptly alert the astronomy community about interesting observations, allowing researchers around the world to turn their telescopes toward the detected phenomena at the right time. NCSA provides data archive To lay the foundation for the LSST cyberenvironment, NCSA and NOAO are developing a prototype data pipeline using the vast stores of data generated by the ground-based observatories NOAO oversees: Kitt Peak in Arizona, Cerro Tololo Inter-American in Chile, and the Gemini Science Center, with observatories in Chile and Hawaii. NCSA created a mirror of the NOAO data archive in Urbana. The archive replication system is built on the Storage Resource Broker (SRB) middleware developed at the San Diego Supercomputer Center (SDSC) and a transfer queuing system NCSA developed for its archive of data from the Berkeley-Illinois-Maryland Array (BIMA) radio telescope. Mirroring NOAO's archive not only gives astronomers a high-bandwidth site from which to access data, it is also part of a strategy of security through redundancy, ensuring that data will survive at one site even if a catastrophic event hits another site. NCSA's robust mass storage system archives more than 2.3 petabytes of researchers' data, with data added at a rate of 40 to 60 terabytes each month. Michelle Butler, the manager of NCSA's Storage Enabling Technologies group, says the data management expertise of the center's staff-encompassing experience with running large file systems, parallel file systems, many storage architecture types, HPC storage, and database storage -- is as much of an asset as the storage infrastructure. "NCSA storage is designed with a long-term view," Plante says. "I can go back to data that's nearly as old as NCSA and have confidence that it will be there." Archiving the data at NCSA also will enable astronomers to take advantage of the center's 41 teraflops of high-performance computing power to process the data. Raw observatory data is marred by artifacts of the observing site, the weather conditions, and the instrument itself; it's as though each image is blurred by fingerprints that need to be wiped away so astronomers can get a sharp view of the data. "We want to get a picture as if it were taken in space, so we have to take out the effects of the telescope and take out the effects of the atmosphere," explains Plante. Tools for accessing data As challenging as the tasks of capturing, moving, processing, and storing data are, they are just the preliminary steps. The real excitement begins when astronomers can access and analyze the data. For LSST, community access will be provided through a Web-based virtual observatory (VO). Therefore, NOAO and NCSA are working to develop the VO model and VO tools, including an authentication and authorization framework for the NOAO portal, an online tool that enables users to find, access, and analyze the data available through multiple public archives, such as the Sloan Digital Sky Survey, Canadian Network for Observational Cosmology, Chandra X-ray Observatory Center, and others. "The NCSA/NOAO collaborative effort provides a backbone for secure data access, which is a vital component for astronomical portals and multi-location image archives," says Chris Miller, an assistant astronomer at Cerro Tololo Inter-American Observatory and one of the collaborators on the NOAO portal project. "These security measures are currently missing from ... most astronomical archive tools and services." By leveraging grid technologies -- including the Globus Toolkit, NCSA's MyProxy, and PURSe (Portal-based User Registration System) -- NCSA and developers from both Argonne National Laboratory and the National Virtual Observatory project have simplified the authentication and authorization process for users. When users log in, the portal knows which services and data they can access and which are withheld from them. This prototype portal was demonstrated in January at the American Astronomical Society Meeting in Washington, D.C. "Our demo included the ability for users to register and log on so that only they could see their proprietary data," Miller explains. "The next version, scheduled to be complete by July, will have a fully implemented security model developed by NCSA in collaboration with the NOAO." The grid-based security framework will serve as an international standard for virtual observatory interoperability. "The development of this security model is fully within the standards and procedures defined by the International Virtual Observatory Alliance," Miller says. "Thus, led by Ray Plante and Ramon Williamson at NCSA, and in collaboration with the NOAO Data Products Program Team based in Chile, we are leading the nation in applying a standard user authentication and authorization model to astronomical portals, archives, and analysis services." Testing the prototype pipeline In July 2006, the LSST project began its first Data Challenge, a test designed to evaluate the development of the data pipeline. The annual challenges will provide feedback for the team and will help the collaborators further refine the requirements of the LSST pipeline. When the observatory begins operation in 2013, data will move from the telescope to a nearby base camp, where some limited data processing will take place in order to provide feedback to the telescope and rapid alerts to the astronomy community. From the base camp, raw data will be transmitted to archive centers for processing, storage, and dissemination to astronomy researchers. Three sites in the National Science Foundation-funded TeraGrid network are standing in for the telescope (Texas Advanced Computing Center), the base camp (San Diego Supercomputer Center), and the archiving center (NCSA). Data will be transferred from site to site and processed along the way in order to evaluate the design of the prototype data management system. This prototype integrates grid technologies with components developed by NCSA's partners at the LSST Corporation, the National Optical Astronomy Observatory (NOAO), the Stanford Linear Accelerator Center, and the University of Washington. First, the challenge will test the data replication software, which is used to transfer data from site to site. Developed at NOAO, the Data Service (DS) software leverages SDSC's Storage Resource Broker (SRB) software. Then the basic functionality proposed for the data processing pipeline will be evaluated using prototype science codes and "resource consumers" that model how actual algorithms would consume compute cycles. These codes will be stitched together through middleware components developed by NCSA and its partners, to mimic the actions and applications that will be components of the final pipeline. "The challenge mimics the data transport and processing as it will happen in real life once the telescope is operating," says Cristina Beldica, project manager for NCSA's LSST effort. "What we're striving to do is get a big enough volume of data to show that our system will scale to the unprecedented data rates we'll see when the telescope is operational." This research is supported by the National Science Foundation. The effort to build the LSST is overseen by the LSST Corporation. For more information, visit its Web site.