Grids for Complex Problem Solving , 29 January 2003

Grid based collaborative working in large distributed organisations

guillaume.alleon@eads.net

Grids for Complex Problem Solving , 29 January 2003

• Overview• Existing Technologies• The future

Outline

Grids for Complex Problem Solving , 29 January 2003

Overview

Grids for Complex Problem Solving , 29 January 2003

Industrial Challenges

Competitive pressure

Time-to-market

ROI focus

More computing power

Multi-disciplinary collaboration

Effective quality & cost control

Grids for Complex Problem Solving , 29 January 2003

Resources in an international company

DE

UK

FR

Projects or VOUsers

Resources

Grids for Complex Problem Solving , 29 January 2003

DE

UK

FR

Over loaded

No easy accessUnknown by user

Non optimized resources utilization

Grids for Complex Problem Solving , 29 January 2003

Grid Middleware

DE

UKFR

UKFR

DE

Partners

Optimized resources usage

Grids for Complex Problem Solving , 29 January 2003

Expected benefits from the systems

• Modelling & Simulation from anywhere, at any time

• Flexible computing power• Better collaboration between engineers • Ressources utilization accounting• Improve productivity• Shorten Time-To-Market• Maximize IT return on assets

Grids for Complex Problem Solving , 29 January 2003

Technologies

Grids for Complex Problem Solving , 29 January 2003

Grids for Complex Problem Solving , 29 January 2003

What is the future ...

Grids for Complex Problem Solving , 29 January 2003

Collaborative working ...

Grids for Complex Problem Solving , 29 January 2003

• Networks– need good connectivity at “low” price

– QoS is a mandatory feature (need to dedicate BW according to criticality of data streams i.e. video or RT needs)

• Grid Middleware– security is probably the most important issue in the

context of VOs

– but directory services, data management, scheduling and brokering are obviously important (eBusiness)

Anatomy of the system ...

Grids for Complex Problem Solving , 29 January 2003

• Applications– at the client level: need for light client (mobile

users)

– applications coupling (soft/soft or soft/hard) requires some specific middleware

• data-fusion

• data-mining

– powerful GUI• 3D RT capabilities

• multi-modal GUI

Anatomy of the system ...

Grids for Complex Problem Solving , 29 January 2003

• Current grids are static, need to be more dynamic– virtual organization can be O(2) to O(1000)

– lifetime from minutes to years

– admin policies shall cross the domains

– dynamic behavior (membership, status, …)

– scalable in term of administration

– resource discovery

– innovative license management policy

Grids for Complex Problem Solving , 29 January 2003

• Security is a MAJOR concern

– how to efficiently protect our data while granting specific access to our resources to providers/suppliers ?

– need to have a fine grain access policy while granting access to community that can be large

Grids for Complex Problem Solving , 29 January 2003

“Research” issues

Grids for Complex Problem Solving , 29 January 2003

– Manageable (interoperable/scalable) security solutions for VOs

– Need for a set of standard industrial applications (e.g. CAE, PDM, 3D imagery) to be “gridified” (vertically integrated) with grids (the foundations are there, but the applications aren’t using them!)

– Need for generic enabling VO services such as community accounting & billing services and industrial strength basic collaboration services

Grids for Complex Problem Solving , 29 January 2003

– WANTED: distributed operating systems “à la” Mosix or Kerrighed. A “Grid Virtual Machine” providing a compact & common set of protocols !

– Convergence of Grid computing & cycle stealing approach (more than P2P) in order to accommodate high performance collaborative applications

– Integration of Knowledge Based Engineering tools within generic/specific Grid based Problem Solving Environments

Grids for Complex Problem Solving , 29 January 2003

Questions ?

Is your company already

using Grid technologies ?

Are your design processes collaborative

& made of multiple participants ?

What do you see as the main

barrier for using Grid computing ?