Summary of DS3 workshop at VOTech kick-off meeting

Single Sign On: Shibboleth & PERMIS

Shibboleth:

• Does it let us chain resources?
• Does is support the VO idea of community?
• How well does it support group membership?
• Can we require service providers to run it? Do we need to?
• Does the anonymizing function break things? (Probably not)

PERMIS:

• How well can it represent VO permissions?
• If we use Shibboleth, must we use PERMIS?
• Is it worth having PERMIS without Shibboleth?

Single Sign On: other issues

Relationship of VOTech work to IVOA discussions:

• Expect IVOA to set the standard
• Should VOTech propose a complete standard to IVOA? (IVOA has an SSO draft already)

Gateways:

• Look at a gateway to translate full SSO to local username-password system => easier for service providers
• Investigate KQML as attribute-passing mechanism

Interoperability of archives

How to get higher-order functions in e.g. queries?

• Open question
• General feeling that “broker” services needed “above” DSA layer

Agents: questions raised

Take-up in astronomy?

• Mainly for robotic telescopes
• Other uses planned, not proven yet

Main benefit to VOtech?

• Autonomy of programmes/processed
• => increased flexibility, robustness

Do we need agents embedded in data centres?

• Probably not.
• Initially prefer agents in broker layer; need not be provided by D/C operators

Do agent P2P techniques disrupt client-server model current in VO?

• Probably not.
• Disruption could be combined to broker layer
• Hierarchy of services might be a problem for agents

Agents: recommendations of workshop

Develop cost-benefit analysis for use of agents

• What can we do better with agents (quantify?)
• Development costs
• Disruption costs

Concentrate on agents as advisers and brokers

Special case: check costs of connecting VO to telescopes via existing eSTAR

• Possible cheap, quick win

CEA: options for standardization

We may wish to adopt AstroGrid's Common Execution Architecture as a standard. There are three ways this could be done.

1. VOTech to IVOA
   • AstroGrid proposes to VOTech
   • VOTech develops
   • VOtech proposes to IVOA
2. VOTech local standard
   • Use inside VOTech-affiliated projects
   • Other arrangements for rest of IVOA
3. Straight to IVOA
   • AstroGrid proposes to IVOA
   • IVOA standardizes
   • VOTech adopts IVOA standard

CEA recommendations of workshop

AVO January-2005 demo includes CEA

• Wait for demo
• Decide whether VOTech likes CEA after demo

Workflow is separate from CEA

• Promote CEA separately (quicker debate)

Archive architecture

Layered architecture is a good thing.

* Get DCA to promote this? * Assume layered architecture in DS3 experiments/prototypes? * DS3 products will mainly be in translation layer.

Grid involvement

How do we embrace the grid?

• Open question
• Minimal cost seems to be few staff-months for 1st application
• Grids are getting easier. Wait for developments?

Recommendation: no grid research in VOTech

• We don’t develop new grid techniques
• We may use existing grid techniques

Interaction with other VOTech groups

• Many things need prototyping.
• Most of these start with other DS* groups, especially User Tools, Resource Discovery and Visualization/Data Mining.
• Those prototypes will reveal infrastructure requirements.
• We need the other groups to feed back requirements into DS3.

-- GuyRixon - 19 Nov 2004