:: Session 2: Day 2: Time 11:00 to 12:30
Principles of Distributed and Grid Systems

Professor Malcolm Atkinson (Programme Chair)

Contents:
11:00 Distributed Systems Architectures & Principles
Presented through a sequence of composable system design patterns
Client-server model – their variations - examples
Central Server models – their limits
Central-server examples - examples

The Service Registry design pattern
   Information about available services, resources, …
   Updated when changes occur
   Interrogated to find services & resources
Challenges for Registries
   Agreeing the descriptive terms
   Matching the terms
   Keeping information sufficiently up to date
   Scalability
Examples you will meet in the ISSGC’06

The Gateway design pattern
   Hiding detail
   Regulating access
   Permitting local control
   Reducing exposed complexity
   Presenting a uniform interface that hides resource differences
Challenges for gateways
   Delivering throughput
   Balancing provider and user requirements
   Designing and delivering fair policies
   Providing external information about status and progress
   Recovering useful work from partially completed requests
   Avoiding persistent loss of allocated resources
Examples you will meet in ISSGC’06

The Master-slave design pattern
   Higher-level composite requests to master
   Master breaks down task into subtasks and allocates to slaves
   Master observes slaves and deals with their progress or lack of it
   Master assembles and reports progress information
Challenges for Master-slaves
   Describing tasks in high-level terms
   Defining and implementing fair policies
   Describing progress
   Scalability and Reliability – handling failures
Examples you will meet in the ISSGC’06

Replication patterns
   For scalability & reliability
   Redirection and allocation algorithms
   State distribution algorithms
   Recovery algorithms
Challenges for Replication
   Locating & choosing a replica
   Planning the distribution of replicas
   Managing change
Examples you will meet in the ISSGC’06

Peer-to-peer design patterns
   Dynamic choice of the allocation of roles
   Algorithms to distribute and manage work dynamically
   Dynamic progress analysis and repair strategies
Challenges for P2P
   Design of stable and efficient algorithms
   Diagnosis of failures and repair strategies
Examples you will meet in the ISSGC’06

Relationship between web services and grids
   Trading between regularity and autonomy
   Management, Development and Deployment
   The role of standards
Examples you will meet in the ISSGC’06

Dynamic composition and bindings
   Enabling better abstraction
   Accommodating incremental change
   Meta data driving the compositions and bindings
Examples you will meet in the ISSGC’06


Slides:
- Distributed Systems: Introduction, Principles & Foundations [ppt | pdf]


Biographies:

Professor Malcolm Atkinson PhD, FBCS, FRSE

Malcolm Atkinson is the Director of the National e-Science Centre and the e-Science Institute. He is the UK e-Science Envoy and plays a leading role in OMII-UK, and is on the advisory boards of GOSC, NCeSS, Baltic Grid and GEON. He leads training and education in the two EU-funded projects EGEE and ICEAGE project, International Collaboration to Extend and Advance Grid Education. These two projects have organised the ISSGC06. He is a member of the Global Grid Forum Steering Group and Data Area Director for GGF.

He began his career in computing in 1966. He has worked at seven universities: Glasgow, Pennsylvania, Edinburgh, UEA, Cambridge, Rangoon and Lancaster; and for two companies: Sun Microsystems (at SunLabs in California) and O2 (an Object-Oriented DB company in its early years in Versailles). He led the development of the Department of Computing Science in Glasgow and is now Professor of e-Science in the School of Informatics, University of Edinburgh. He has more than 130 publications. He has taken leading roles in national strategic research and infrastructure committees.

website!