Construction of the Japanese Virtual Observatory JVO

1
Construction of the Japanese Virtual Observatory
(JVO)
Yoshihiko MIZUMOTO, Masatoshi OHISHI, Naoki
YASUDA, Yuji SHIRASAKI, Masahiro TANAKA (NAOJ),
Yoshifumi MASUNAGA (Ochanomizu Univ. and NAOJ),
Ken MIURA, Hirokuni MONZEN, Kenji KAWARAI,
Yasuhide ISHIHARA, Yasushi YAMAGUCHI and Hiroshi
YANAKA (Fujitsu Ltd.) Contact Address
adac_at_jvo.nao.ac.jp
Abstract The National Astronomical Observatory
of Japan (NAOJ) has been operating several large
astronomical facilities, such as the SUBARU
telescop, the 45 m radio telescope and the
Nobeyama Millimeter Array, and plans to construct
the ALMA under close collaborations with the US
and the EU. Since January 2002, the NAOJ has been
connected to the SuperSINET with 10 Gbps, and it
has become possible to provide huge amount of
observed multi-color data and analyses facilities
to other astronomical institutions. Thus we have
started the Japanese Virtual Observatory (JVO)
project since April 2002. JVO utilizes the Grid
technology to combine several remote
computational facilities. We have completed to
define the query language for the JVO, and have
been designing on the deployment of JVO
components. We plan to construct a JVO-prototype
by the end of 2002. More information is provided
at

http//jvo.nao.ac.jp
Ultrahigh-speed network and JVO
Schematic diagram of JVO system
Distributed data base system
Users
Inter-national VO
10Gbps Super SINET, linked with GRID technology
GRID
Distributed computing system
The JVO, the distributed data base systems of
Subaru and Nobeyama observatories, and
astronomers in the research institutes are linked
over
JVO _at_ NAOJ
the GRID technology through the 10Gbps Super
SINET (http//www.sinet.ad.jp/english/).
Super SINET is an ultrahigh-speed network
intended to develop and promote Japanese academic
researches by strengthening collaboration among
leading academic research institutes. The
National Institute of Informatics has been
operating the network since January 4, 2002. The
Internet backbone connects research institutes at
10 Gbps and the leading research facilities in
the research institutes are directly connected at
1 Gbps.  

• JVO consists of a distributed computing system
  (DCS) which is deployed over the GRID technology
  and a registry which provides information
  required for DCS to query the distributed DB
  system.
• All the computers of the DCS have an equivalent
  function and any of them can takes place of
  another machine, which is important for
  robustness of JVO system.
• Selection of a machine for servicing the JVO
  users is automatically performed by GRID system
  based on the system load average.
• JVO has inter-operability with the other VOs.

Properties of the astronomical data base utilized
by JVO
Three-Tiered Design of the JVO Prototype
The JVO Query Language
The JVO prototype is now under development. The
design of the JVO prototype is shown as a
schematic diagram. We adopted to use the Globus
Toolkit V2 for our prototype. However we also
take into account the Web service concept which
will be included in the Globus Toolkit V3.
First of all, researchers provide the JVO with
simple instructions how they plan to use their
own ''Virtual Observation''. The JVO portal
interprets them and generates a work flow through
consulting the UDDI servers, where available JVO
services are registered. Based on the work flow,
built-in or user-defined services are called.
The GRID framework is used for dynamical
assignment of distributed resources according to
their availabilities. Execution results of the
work flow are transferred through GridFTP and
presented to the researchers with the skycat, etc.
AREA table
s
t
a1
select s.a, t.a, ... from
SUBARU.R s, 2MASS.K t, ... where
(s.AREA() OVERLAP t.AREA()) as a
Partition to small segment
a2
Future Plan
a3

• Federate with more data
• Data of SUBARU open use/Nobeyama Radio
  Observatory
• Interoperability with other VOs
• Toward International VO
• CPU intensive image analysis tools
• Deconvolution, image subtraction, ... Run on PC
  cluster via GRID
• Data mining / visualization tools
• Manage huge amount of data

cutout request
s.AREA()
t.AREA()
OVERLAP
JVOQL has an ability to query image data without
referring to catalogs. This ability is useful for
multi-color or multi-epoch analyses. The above
JVOQL example shows how to obtain R-band
SUBARU.R s
2MASS.K t
images taken by SUBARU and K-band images by 2MASS
in an area where both SUBARU and 2MASS observed.
The operand OVERLAP returns overlapped area of
the two data. Similarly the operand X.AREA()
returns the observed area of server X.