Home

From SmartGrid

1 Overview
2 SmartGRID Node
3 Smart Resource Management Layer (SRML)
4 MaGate
5 Acknowledgement
6 Papers
7 Events
8 Software Download

Overview

SmartGRID proposes to tackle Grid Scheduling problem by a middleware approach. In opposition to most scheduling and load-balancing systems in network computing, this middleware is modular. It clearly decouples scheduling from network monitoring. Scheduling module aims at serving applications while making the best use of the available computing resources; the role of monitoring is to constitute knowledge about the network and the resources. The projected middleware is then structured in two layers: Smart Signaling Layer (SSL) is in charge of monitoring the computing network; while Smart Resource Management Layer (SRML) is responsible for dynamic scheduling.

SmartGRID Node

SmartGRID aims at constructing an integrated high-level grid community. The grid community is constituted of all the logically connected and engaged SmartGRID Nodes (SG-Node). The SmartGRID Nodes can behave different roles depending on their positions within the grid community.

To achieve our design goal with SmartGRID layered architecture, each individual SG-Node is comprised of three entities, each entity belonging to a separated layer:

The MaGate from SRML and acts as the gateway of the SG-Node.
The Nest from SSL and continuously constitutes the knowledge of network and resources.
The DataStorage from DWI and fills the communication gap between SRML MaGate and SSL Nest.

Smart Resource Management Layer (SRML)

The Smart Resource Management Layer (SRML) is designed and implemented in EIA-FR, which aims at serving grid application with dynamic scheduling methods, working on the basis of information provided by the ant layer. Information will be exchanged through interface data warehouses.

MaGate

The Smart Resource Management Layer (SRML) consists of the SmartGRID high-level schedulers named MaGate. MaGates are modular, and aim at being an open structure, fully decentralized, and interoperable high-level schedulers, capable of integrating diverse external grid components easily. The core value of MaGate is interoperability. Beyond the high-level scheduler’s traditional functionality of dispatching scheduling decisions to low-level resource management systems or middlewares, MaGates focus on the interoperability of high-level schedulers, which facilitates negotiation and transport scheduling events (requests, responses) amongst diverse virtual organizations and grid systems. In this context, MaGates empower the users to utilize resources outside the already known grid systems, with fully decentralized topology and high reliability. The continuously updated and reliable information within DataStorages contributed via the external components, especially SSL Nests, provides the grid infrastructure information as the basis of MaGates’s interaction. Furthermore, the MaGates’ interoperation is based on several existing and merging protocols and standards, such as WS-Agreement, JSDL and SDL.

To achieve the aforementioned design purpose, the MaGate comprises five separated modules: the Kernel, the Interface, the Community, the LRM, and the External, for increasing system flexibility and compatibility. The modules are not monolithic and each component’s involved services are also configurable.

Acknowledgement

SmartGRID project is developed thanks to the financial support of the Swiss Hasler Foundation, in the framework of the ‘ManCom Initiative’ (ManCom for “Managing Complexity of Information and Communication Systems”), project Nr. 2122.

Papers

Recommended:

Ye Huang, Nik Bessis, Peter Norrington, Pierre Kuonen, Beat Hirsbrunner, “Exploring decentralized dynamic scheduling for grids and clouds using the Community-Aware Scheduling Algorithm”, Future Generation Computer Systems (FGCS), Elsevier, 2011. DOI:10.1016/j.future.2011.05.006 {Accepted}

Nik Bessis, Ye Huang, Peter Norrington, Antony Brown, Pierre Kuonen, Beat Hirsbrunner, "Modelling of a Self-led Critical Friend Topology in Inter-cooperative Grid Communities", in: Simulation Modelling Practice and Theory (SMPT), vol. 19, nr. 1, pp.5-16, Elsevier, 2011.

For the full publication list, please refer to:

Publications