Presentation on theme: "1 Project Overview EconomyGrid Economic Paradigm For “Resource Management and Scheduling” for Service-Oriented Grid Computing Presenter Name: Sama GovindaRamanujam."— Presentation transcript:
1 Project Overview EconomyGrid Economic Paradigm For “Resource Management and Scheduling” for Service-Oriented Grid Computing Presenter Name: Sama GovindaRamanujam
2 What is the “Grid” ? An infrastructure that tightly integrates computations devices, software, catalogued databases, specialized instruments, displays and people from widespread locations, and under different management authorities. Motivation To address the drivers for HPC, which are notably areas like life sciences, digital imaging, CAD/CAM, e- commerce, aerospace and military applications.
4 Challenges Security (authentication, authorization) etc., resource allocation, scheduling, access of remote data sets or even collaboration on results. Research initiatives Globus A joint initiative between the Argonne National lab, NASA, American national labs, and numerous universities.
8 Economy Grid The Team Dr. Rajkumar Buyya Dr. David Abramson Dr. Jonathan Giddy
9 Focus Of Research The development of economic or market-based resource management and scheduling system for global grid computing. Research Streams Of Ecogrid (broadly) GRACE GRid Architecture for Computational Economy GRB Grid Resource Broker (Super Scheduler/MetaScheduler)
10 GRACE Supports generic interfaces (APIs) that can be used by the grid tools and applications programmers to develop software supporting the computational economy. Key components Grid Resource Broker (e.g., Nimrod/G) Grid Resource and Market Information Server (provided by Globus and extended services for resource access price) Grid Open Trading Protocols and API Trade Manager (part of broker involved in establishing service price) Grid Trade Server (works for Grid Service Providers)
11 Grid Resource Broker (GRB) It is a grid application scheduler Responsible for resource discovery, selection, and scheduling computations over them. Computational model of applications to be scheduled on Grid can be sequential or parallel in nature. Applications can be parameterized using a Parameter Sweep Specification (PSP) language.PSP To perform parametric executions, the user simply specifies input parameters and commands to be executed. The PSP does the rest - it generates jobs to be computed, executes the jobs and collects the results. Eg:-enFuzion by TurboLinux
12 Application: “Molecular Modeling for Drug Design"
13 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. Nimrod/G resource broker Framing Engine provides protocols and APIs for developing Nimrod/G clients for computational steering (e.g., Nimrod-G Monitor), customized applications for parametric computing (e.g., Active Sheets), and implementing user-level job schedulers.Nimrod-G MonitorActive Sheets Variations Of Nimrod Nimrod-G Nimrod-L Nimrod-O
16 Economic Models for Resource Trading Commodity Market Model Posted Prices Model Tendering Model Auction Model Shareholder Model Partnership Model
17 Commodity Market Model
18 Posted Price Model
19 Tender Model
20 Auction Model
21 Shareholder/Partnership Model
22 GridSim- A Grid Simulation Toolkit for Resource Modeling and Application Scheduling for Parallel and Distributed Computing The primary objective of the project is to investigate effective resource allocation techniques based on computational economy through simulation. The GridSim toolkit will be released soon along with source code under GPL-like license.
23 Some Quotes ".NET is grid...the grid is quite interesting because it's the ultimate in distributed computing.“ says Microsoft researcher Gordon Bell "The holy grail of the grid is to make computers like the telephone system you can dial into a phone anywhere in the world and talk as fast as you like." says Appelbe, Chief executive officer of the Victorian Partnership for Advanced Computing (VPAC) "At the end of the day, the critical question is one of latency when you're talking about network grids." says Professor Greg Egan, Dir of the Centre for Telecommunications/Information Engg,Monash Univ.