Outline ICRAR MRO MWA IVEC NGAS GAVO Our

中国科学院新疆天文台 Outline: ICRAR, MRO, MWA, IVEC NGAS GAVO Our DC

中国科学院新疆天文台 ICRAR The International Centre for Radio Astronomy Research is a collaborative centre based in Perth, Western Australia. Already achieving research excellence in astronomical science and engineering, ICRAR has grown steadily since its launch in September 2009. MRO The Murchison Radio-astronomy Observatory (MRO) is the home of the Australian SKA Pathfinder (ASKAP) and the Murchison Widefield Array (MWA). It will shortly also be the home of a significant part of the Square Kilometre Array (SKA) itself (South Africa will also host part of the SKA). It is located in Western Australia’s Mid West, approximately 300 km north east of the coastal city of Geraldton, within the Shire of Murchison. This outback region is roughly the size of the Netherlands with a resident population of around 140 people and a population density of around one person for every 350 sq. km. The centre of the 70 km radius circle defining the MRO is at S 26 o 42’ 15”, E 116 o 39’ 32”. The MRO and the surrounding landscape are protected through federal legislation from any activities that might interfere with the operation of radio telescopes within 150 km of the core site. The establishment of this ‘Radio Quiet Zone’ (RQZ) combined with a very low and sparse local population means the MRO is the perfect place for constructing and operating the most sensitive radio telescopes the world has ever seen.

中国科学院新疆天文台 MWA The MWA consists of 2048 dual-polarization dipole antennas optimized for the 80 -300 MHz frequency range, arranged as 128 "tiles", each a 4 x 4 array of dipoles. The array has no moving parts, and all telescope functions including pointing are performed by electronic manipulation of dipole signals, each of which contains information from nearly four steradians of sky centered on the zenith. Each tile performs an analog beamforming operation, narrowing the field of view to a fully steerable 25 degrees at 150 MHz. The majority of the tiles (112) are scattered across a roughly 1. 5 km core region, forming an array with very high imaging quality, and a field of view of several hundred square degrees at a resolution of several arcminutes. The remaining 16 tiles are placed at locations outside the core, yielding baseline distances of about 3 km to allow higher angular resolution for solar burst measurements.

中国科学院新疆天文台 ASKAP is designed to be a high speed survey instrument with high dynamic range. This requires an extremely wide field-of-view interferometer with Phased Array Feeds (PAFs). The PAF has a total of 188 elements, 94 in each of two polarizations. The specification for ASKAP is: A total collecting area of approximately 4, 000 square metres, from 36 antennas each 12 metres in diameter System temperature less than 50 K Frequency range from 700 MHz to 1. 8 GHz 300 MHz instantaneous bandwidth 36 independent beams, each of about 1 square degree, yield overlapping to a 30 square degree field-of-view at 1. 4 GHz Maximum baseline of approximately 6 km Full cross-correlation of all antennas Possible remote array station capability located in NSW, approximately 3, 000 km from the core site.

中国科学院新疆天文台 IVEC is an unincorporated joint venture between CSIRO, Curtin University, Edith Cowan University, Murdoch University and the University of Western Australia and is supported by the Western Australian Government. Fornax The SGI supercomputer comprises 96 nodes, each containing two 6 -core Intel Xeon X 5650 CPUs, an NVIDIA Tesla C 2075 GPU and 72 GB RAM, resulting in a system containing 1152 cores and 96 GPUs. Storage Data Stores @ i. VEC also includes the Petabyte Data Store (PB Store) which is a scalable storage solution based on Sun’s SAM-QFS (Storage and Archive Manager File System & Quick File System) software, and consists of a 70 TB of disk cache and a second 100 TB file system. The SAM-QFS software automates the migration and retrieval of files across multiple tiers of storage from the online disk cache, to near-line tape storage, to offline tape archival. This system allows researchers to see their complete directory hierarchy and have seamless access to their files within a single file system. Each file that is placed on the system has a copy written to two separate tapes. LTO 4 and Storage. Tek SL 8500

中国科学院新疆天文台 NGAS (Next Generation Archiving System) ESO’s answer to the data explosion challenge. It is an integrated system consisting of hardware and software. Andreas wicenec andreas. wicenec@uwa. edu. au Research Winthrop Professor UWA data intensive research expert. Specialisation: Data flow and storage design and HPC integration

中国科学院新疆天文台 NGAS, NG/AMS, NGAST? ? • • Next Generation Archive System: ESO’s answer to the data explosion challenge. It is an integrated system consisting of hardware and software. NG/AMS: Next Generation Archive Management System; this is the NGAS software. NGAST: The name of the ESO project leading to the NGAS. Started in 2000. First prototype system installed on La Silla in 2001. Now there are >100 NGAS servers in the ESO Archive. Never call something NG? True, but the intention is to keep this at the forefront of technology. 14

中国科学院新疆天文台 Scope of NGAS n Secure, scalable, distributed and affordable file store. n Secure: Do not lose files. n Scalable: Be prepared for peta-bytes of data. n Distributed: Operate transparently across physically and logically separated locations. n Affordable: TCO at least comparable to DVDs in juke-boxes. n Lowest component of the archive. 15

中国科学院新疆天文台 NGAS concepts n Distributed storage, but not global file system. n Almost complete separation of storage and business logic: NGAS stores many files on many media, which may be distributed and moved around the globe. ¨ Knowledge and functionality to extract the contents, do groupings and associations is somewhere else. ¨ n NGAS deals only with file names, IDs, sizes, locations, media IDs, capacity…. And that information is stored in a data base. 16

中国科学院新疆天文台 NGAS concepts n No user knowledge required about the physical location of a given file. In an operational environment there is one central access point (master), which passes requests to the correct NGAS server using HTTP proxy or HTTP redirect. Any server can be used as a master. n Ensure that there always at least two copies on two different media of any given file in the system. » Special commands for file and media removal. n Active health check of media and files. n Active consistency checks of media contents vs. DB. n NG/AMS uses modular plug-in architecture. » Very flexible and easy adjustments to hardware and environment changes. 17

中国科学院新疆天文台 Advanced NGAS features • Subscription service: –One NGAS server can subscribe to another. Parallel cluster mirroring. • Disk and file cloning. • Processing plug-in. • Power saving mode: Data flow NGAS Archiving Subscription NGAS • NGAS –Automatic sleep and wakeup system built in. 18

中国科学院新疆天文台 NG/AMS Interfaces n All user I/F and APIs are based on HTTP. n Management information is kept in a DB. n Interfaces for five DBs. n Interface to syslog n E-mail notification services. n Message and config-file format is XML. 19

中国科学院新疆天文台 NGAS as an Embedded System n NGAS is a tightly integrated system in order to be ready for operations: ¨ Minimal boot time ¨ Automated boot and shutdown procedures ¨ Hardware monitoring ¨ Integration of raid controller ¨ Media verification using access to on-disk BIOS ¨ Cluster configuration Virtual. Env 20

中国科学院新疆天文台 Implementation n 100% Python n about 25, 000 lines + 15, 000 lines of in-line documentation n advanced system integration tests n two person team, plus some help from system admin people. n additional code added for mirroring n Advanced fully automatic installation procedure.

中国科学院新疆天文台 Data Intensive Research Pathfinder Hierarchical Storage Manager

中国科学院新疆天文台 What’s Unique? n Transparent global distribution of archival data. n Global mirroring service over slow networks n Fully traceable disk based data transfer n Very robust and flexible (plugins) n Full control of code n Used to store VLT, ALMA, MWA, VLA, e. VLA and VLBI data. n Supports processing inside archive

中国科学院新疆天文台 Where Does it Fit? 40 T MRO MWA IVEC-DC Raw Visibilities 1 Gb/s 80 T Raw Visibilities 10 Gb/s MRO MWA IVEC-DC

中国科学院新疆天文台 GAVO (The German Astrophysical Virtual Observatory (GAVO) Dr. Markus Demleitner Astronomisches Rechen-Institut in Heidelberg. Woring on the GAVO project, an effort to contribute to Virtual Observatory, e. g. , through the operation of the GAVO data center and developing the Da. CHS multiprotocol server suite. http: //www. tfiu. de/homepage/ gavo@ari. uni-heidelberg. de

中国科学院新疆天文台 GAVO (The German Astrophysical Virtual Observatory (GAVO) GAVO is the German contribution to the IVOA (International Virtual Observatory Association), the international effort to create and expand the Virtual Observatory (VO). ) The goals of the Virtual Observatory are to allow or improve access to astronomical data of all kinds (astrometry, photometry, spectroscopy, time series, . . . ), from everywhere, in well-defined protocols, let astronomers easily discover, access and use data relevant to them, ensure that data does not simply disappear, that it is properly described and can be accessed and understood in the future, provide software to help astronomers to use all this.

中国科学院新疆天文台 Tier 0 Online processing Online archive Correlator 25 M PFB Nan Shan Receiver 10 Gb/s NGAS Client Data Capture 10 Gb/s Tier 2 Tier 1 1 G/10 Gbps Mirrored Archive (All Data) Publish Data 1 Gbps Dedicated Network Qi Tai Site Nan Shan Other Places MS Backup Q A Raw Data Long-term archive RS RV XAO Headquarters (Storage Arrays) M S local storage Computing resources Processed Data VO Catalog Data Query Web Scripting HPC VO Based WEB Interface UI XAO’s Future Data Archiving, Processing and Query System Image, Time sequence Zhanghailong

中国科学院新疆天文台 ICRAR NGAS MRO GAVO MWA OUR DC Thank you Question? 38