Wednesday, February 23, 2005

Infrastructure

Today we're talking about infrastructure (DS3), which is the bit of the meeting in which I'm probably most interested...


Why does it always snow at VOTech meetings?

DS3 is about enabling science rather than actually doing science, so the vital issues are service availablity, reliability and scalability along with usability, support and interoperability.

The first presentation of the meeting was from Guy Rixon who talked about security for the VO. The services we run should not compromise the host, the services we install should run with minimal privileges and with minimal third party software needs. Additionally any security we impose must be easy to use and relatively transparent to the user. In effect this means we need single sign on, and this has far reaching implications.

Noel Winstanley then talked about enabling users to actually do science.


Noel talking about interoperability

He talked about how the VO should be language neutral, and accessible from the existing tools the users are currently using. Amongst various ideas to achieve this his suggested solution is a proxy application running on the user's desktop, which he is calling Dashboard, which uses HTTP and XML-RPC to expose various services to the user in a consistent way. This actually looks like a good solution to the various interoperability problems, and means that you can access quite a lot of the VO, that isn't currently accessible, programmatically from your favourite language.

Then just before lunch Thomas Boch talked about the proposals from CDS about what DS3 should be doing with its time. Amongst other things, but crucially from the eSTAR perspecitive, they are anticipating offering proper SOAP access into VizieR in the near future. Interestingly, they're also proposing a graphical workflow creation tool, this was talked about during the AstroGrid rollout back in December as an alternative to the current portal.

Discussions after lunch started with Paul Harrison talking about CEA, what it is and why it's useful. It was suggested that CEA was refactored in terms of WS-RF which would make it more accessible and language neutral, since there is a decent WS-RF implementation for Perl, this would suit me just fine.

Finally we talked about what should be the priorites for DS3, where we are now and what we should be doing next...