Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.adass.org/adass/proceedings/adass96/reprints/chavana.pdf Дата изменения: Wed Jan 14 23:15:27 1998 Дата индексирования: Tue Oct 2 12:43:56 2012 Кодировка: Поисковые слова: п п п р п

Astronomical Data Analysis Software and Systems VI ASP Conference Series, Vol. 125, 1997 Gareth Hunt and H. E. Payne, eds.

A Distributed System for "Phase I I" Proposal Preparation
A. M. Chavan and M. A. Albrecht European Southern Observatory, Karl-Schwarzschild-Str. 2, D­85748 Garching, Germany; E-mail: amchavan@eso.org Abstract. A significant fraction of observing time on ESO's new Very Large Telescop e will b e sp ent in service mode; therefore, investigators need to describ e in detail what they want to observe, and how. ESO is prototyping a distributed "Phase I I" system that will allow astronomers to prepare their observations at their home institutions, while maintaining a central rep ository for all observation data. Astronomers will b e offered a wide array of observation preparation tools, including data entry GUIs, instrument simulators, and catalog interfaces. The system is structured as a set of distributed clients and centralized servers, and it op erates as a front-end to the other elements of the VLT Data Flow System.

1.

Introduction

In order to maximize the scientific throughput of the system, a complex and exp ensive observatory like ESO's VLT should let the most demanding observing programmes take advantage of the very b est weather conditions. Traditionally, this has not b een p ossible: observers have had to cop e with the atmospheric conditions prevailing during the night assigned to them. ESO is exp erimenting with op erating telescop es in service observing mode, whereby staff observers execute scientific programmes on b ehalf of the investigators (this mode of op eration is also called queue observing). To make service mode observing p ossible, scientific programmes must b e describ ed in great detail--a process commonly defined as Phase II prop osal preparation. ESO is developing a Phase I I Prop osal Preparation system (P2PP) to collect detailed scientific programme descriptions, in the form of observation blocks (OB). The system is initially directed to supp ort service mode observing (that is, a significant fraction of the VLT observations), but will eventually b e used by "classical mode" observers as well: OBs will b e the main input to the VLT's Data Flow System (GrosbЬl & Peron 1997).

2.

Observation Blocks

VLT observing programmes will b e comp osed of several observation blocks, each coupling a target (what should b e observed) with a description of how the observation itself should b e carried on; observation blocks represent a high level view of VLT op erations. Several scheduling constraints may b e sp ecified for OBs, like 367

© Copyright 1997 Astronomical Society of the Pacific. All rights reserved.


368

Chavan and Albrecht
ObservationBlocks TargetPackages

ObservationDescriptions

OD1

OB1 OB2

TP1 TP2 TP3

OD2 OD3

OB3 OB4

Figure 1. Targets TP1 and TP1 are observed in the same way, since observation blocks OB1 and OB2 share the same observation description, OD1. Conversely, TP3 is shared by two observation blocks, OB3 and OB4. The investigator wishes to observe the same target in two different ways: e.g., with an optical imager and with a sp ectrograph.

worst acceptable seeing, absolute timing for transient phenomena, chaining of observations, maximum acceptable airmass, etc. Targets (technically target packages) are describ ed in terms of coordinates, prop er motion, magnitude, color, and all other information which may b e needed to acquire them and track them during exp osure. This includes one or more guide stars. The observation description is comp osed of series of standard op erations, called templates: typical templates include target acquisition, science exp osures, and calibration frames. For each template to b e executed, astronomers need to sp ecify a set of parameters, like filter names, exp osure times, CCD readout mode, etc. These parameters sp ecify a full instrumental configuration as well as execution details, like the pattern of images for a mosaic template. Targets and observation descriptions can b e shared among OBs, as shown in Figure 1, thus allowing astronomers to easily group related observations.

3.

A Distributed System

For privacy and reliability of information, observation blocks will b e stored in a central rep ository at ESO. Other imp ortant services, like instrument simulators, data servers (Albrecht, Brighton, & Herlin 1997), and help pages, will b e provided centrally by ESO as well. However, it is imp erative that (a) P2PP can take advantage of the investigators' own computing facilities, and (b) astronomers can prepare their observations even when they are not connected to the Internet. These requirements dictate the design of a distributed, client-server system: "lightweight" clients interact with central servers, and can maintain a local cache of information.


A Distributed System for "Phase I I" Prop osal Preparation
ESO Remote host
P2PP Main panel Scheduling server Exposure time server P2PP archive server Template call validation server Target definition GUI

369

...other panels...

...other servers...

Observation description GUI

Template parameter GUI

Figure 2.

P2PP architecture.

3.1.

Architecture

The P2PP system is therefore a collection of indep endent stand-alone modules, structured in a client/server relationship. The server modules run at ESO and handle centralized services, like database access for OB storage, exp osure time calculation, etc., while client modules run on the OB's author's host. Most client modules generate, display, and process part of an OB's information content: there is a module for describing targets, one for observations descriptions, etc. One sp ecial client module is used for global OB op erations, like transfer of OBs to and from the central (ESO) database. All modules offer a graphical user interface. The overall software architecture is shown in Figure 2. P2PP is b eing implemented in C++ and Tcl/Tk; it uses HTTP as the communication protocol b etween clients and servers.

4.

P2PP and the VLT Data Flow System

ESO's Phase I I Prop osal Preparation system is an integral part of the VLT Data Flow System (DFS). Observation blocks are built on the basis of information collected by the Phase I Prop osal Handling and Rep orting System (Chavan 1995), and are fed downstream to the VLT Scheduling Tools (Chavan 1996)--which we are developing in coop eration with the Space Telescop e Science Institute--and to the VLT Control Software. All other subsystems of the DFS--including archiving, pip eline processing, and quality control--deal with observation blocks as well: OBs represent the basic quantum of information flowing through the VLT system.


370 5.

Chavan and Albrecht Planning and Development

The first op erational version of P2PP will supp ort preparation of observations for the ESO's 3.5m New Technology Telescop e, and will b e made available in January 1997. The first VLT version of the system will b e delivered in the spring of 1998. The Web is already an imp ortant enabling technology. In the future, we hop e to b e able to use tools like the Java programming language to provide distributed tools that will run transparently on different hosts. Acknowledgments. We are grateful to the memb ers of the ESO Data Flow Pro ject Team for the long discussions during which the concepts describ ed here were defined. References Albrecht, M., Brighton, A., & Herlin, T. 1997, this volume, 333 Chavan, A. M., & Albrecht, M. A. 1995, in ASP Conf. Ser., Vol. 77, Astronomical Data Analysis Software and Systems IV, ed. R. A. Shaw, H. E. Payne, & J. J. E. Hayes (San Francisco: ASP), 58 Chavan, A. M., Johnston, M., & Albrecht, M. A. 1996, in ASP Conf. Ser., Vol. 87, New Observing Modes for the Next Century, ed. T. A. Boroson, J. K. Davies, & E. I. Robson (San Francisco: ASP) GrosbЬl, P., & Peron, M. 1997, this volume, 22