[My apologies if some of you have already seen this]
As most of you will already know, we have a group sponsored in part by an OPTICON grant which is looking at the issue of future astronomical software environments. This was discussed in a BOF session at the ADASS last fall. Following ADASS the first face-to-face meeting was held at ESO in early December, with the goal of agreeing upon 1) the requirements for such an environment, and 2) a concept for the high level architecture.
We need to do this now as a number of related efforts are starting up at NRAO and ALMA, at ESO, and at other sites, plus we have a need to integrate legacy software with the VO (e.g., within NVO we have a project to interface IRAF to VO, but all such systems face similar problems). While these efforts are all VO-related, the scope is somewhat broader than VO, addressing the general problem of producing compute-intensive astronomical software for pipeline processing as well as for general data analysis.
The December meeting of the OPTICON working group resulted in a preliminary agreement on the architecture required to address this problem, including a strategy for how to integrate this with VO. The intention is to invite broader comment with the goal of agreeing, by the end of January, upon the high level approach to be taken. In particular we need to examine the relationship of this project to VO and answer questions such as: Are the efforts complementary? Are they the same? To what extent do they overlap? The relationship to related efforts within VO such as AstroGrid needs to be examined. In the DAL area, we need something like this to handle the scientific computation part of DAL services (e.g., to produce the virtual data products to be returned), plus we need this on the client side to interface DAL and other VO services to data analysis systems.
The proposed high level architecture from the December meeting is described in the following paper:
http://archive.eso.org/opticon/twiki/pub/Main/GeneralDocuments/CF-Arch.pdf
Currently this is an internal discussion within the OPTICON WG. The above paper is a snapshot of the draft architecture document from the project TWiki which can be found at
http://archive.eso.org/opticon/twiki/bin/view/Main/WebHome
You are invited to review this and comment. It would be good to get at least some preliminary feedback by the time of the IVOA Exec meeting in about two weeks.
A key point here is that, for the most part, our VO efforts have thus far focused exclusively on federation of data archives, and data mining of fully processed data from large surveys. While this is a very important problem, a large part of astronomy is still concerned with PI science where an astronomer applies for time and gets data for some specific science program. In this case the astronomer usually wants to interact directly with "their" data, possibly reprocessing it or performing some custom analysis, and finally performing some sort of multiwavelength analysis where the proprietary data is compared with data from public archives. Hence we deal with pipeline and offline (interactive) processing, with user written software, and with ad-hoc data interaction and analysis. VO is involved mainly in the analysis phase; ideally the astronomer would like to use the same desktop environment for both data processing and analysis. This scenario is central to those of us from observatories which support time allocated PI science, and this is in part what is driving the OPTICON study.