The Grid Economy Project


Economic Paradigm for "Resource Management and Scheduling" for Service-Oriented Grid Computing

Grid Computing Info Centre




Updates:


The focus of our research is on the development of economic or market-based resource management and scheduling system for global grid computing. The research and development is being carried in two threads:

The rationale for using economics as a metaphor for management of resources and scheduling in peer-to-peer and grid computing along with scientific results are discussed in EuroGlobus invited/keynote talk:

GRid Architecture for Computational Economy (GRACE)

We are exploring the development of a generic framework/infrastructure for grid computational economy called GRACE that co-exist with Grid systems such as Globus and Legion for creating an Economy Grid. The key components of GRACE infrastructure include: They work closely with grid middleware and fabrics. The GRACE-infrastructure supports generic interfaces (protocols and APIs) that can be used by the grid tools and applications programmers to develop software supporting the computational economy.

Grid Resource Broker (GRB)

The Nimrod-G Resource Broker is a grid application scheduler and it is responsible for resource discovery, selection, scheduling, and deployment of computations over them. The computational model of applications to be scheduled on Grid can be sequential, embarrassingly or plesantly parallel (task farming), and parallel in nature. Applications can be parameterized (e.g., shop simulation, Drug Design, network and fuzzy logic simulation, high energy physics events processing) using our simple, Parameter Sweep Specification (PSP) language for SPMD (single program, multiple data) style data parallel computing.

We are addressing scheduling of parameter sweep (task farming) applications on World Wide Grid (WWG) resources as part of the Nimrod/G project. The previous version of Nimrod/G supported deadline based scheduling by keeping the cost of computation as low as possible. The latest version supports both deadline(soft realtime) and budget (computational economy) constraits in scheduling and at the same time it can optimise execution time or budget expenses. The GRACE infrastructure will enable Nimrod/G to dynamically trade for grid resources in the open market environment and select resources that meet user requirements (deadline and cost). Our focus is on the the design and development of smart scheduling algorithms that support constraint-based scheduling. Our economics and soft-realtime scheduling algorithms explicity handle: existence of heterogenous and unreliable resources in Grid with varying performance, cost, price, acceess policies, and user constraints such as deadline for completion of the assigned work, and budget limitations. Basically, on behalf of users, broker smartly leases distributed resources to run applications depending on the QoS demands and yet meet user constraints.

Our Nimrod/G resource broker Framing Engine is a persitent and programmable component. It provides protocols and APIs for developing Nimrod/G clients for computational steering (e.g., Nimrod-G Monitor), customised applications for parametric computing (e.g., Active Sheets), and implementing user-level job schedulers.

Grid Simulation (GridSim) Toolkit

Application Initiatives

The Team

The research is carried out by Rajkumar Buyya in collaboration with David Abramson, Jonathan Giddy, and many external collaborators.

The World Wide Grid (WWG) testbed

You are invited to volunteer (idle resources of) your computers to this testbed.

In News and Press

Publications

Presentations

Demonstrations

Other Pointers

Supported by

Collaboration

Globus Project