- DataGrid meeting in Catania – 9 th to 11 th of April, 2001 A pre-alpha accounting architecture The.

Slides:



Advertisements
Similar presentations
CPU Scheduling.
Advertisements

Pricing for Utility-driven Resource Management and Allocation in Clusters Chee Shin Yeo and Rajkumar Buyya Grid Computing and Distributed Systems (GRIDS)
An Efficient Dynamic Auction for Heterogeneous Commodities (Lawrence M.Ausubel - september 2000) Authors: Oren Rigbi Damian Goren.
Priority INHERITANCE PROTOCOLS
Energy-efficient Task Scheduling in Heterogeneous Environment 2013/10/25.
COT 4600 Operating Systems Fall 2009 Dan C. Marinescu Office: HEC 439 B Office hours: Tu-Th 3:00-4:00 PM.
1 Transportation problem The transportation problem seeks the determination of a minimum cost transportation plan for a single commodity from a number.
Silberschatz, Galvin and Gagne  2002 Modified for CSCI 399, Royden, Operating System Concepts Operating Systems Lecture 19 Scheduling IV.
Chapter 5 CPU Scheduling. CPU Scheduling Topics: Basic Concepts Scheduling Criteria Scheduling Algorithms Multiple-Processor Scheduling Real-Time Scheduling.
All Hands Meeting, 2006 Title: Grid Workflow Scheduling in WOSE (Workflow Optimisation Services for e- Science Applications) Authors: Yash Patel, Andrew.
A Grid Resource Broker Supporting Advance Reservations and Benchmark- Based Resource Selection Erik Elmroth and Johan Tordsson Reporter : S.Y.Chen.
Chapter 6: CPU Scheduling. 5.2 Silberschatz, Galvin and Gagne ©2005 Operating System Concepts – 7 th Edition, Feb 2, 2005 Chapter 6: CPU Scheduling Basic.
Economic Principles.  Economics:the study of how human beings allocate scarce resources to produce various commodities and how those commodities are.
Silberschatz, Galvin and Gagne  Operating System Concepts Chapter 6: CPU Scheduling Basic Concepts Scheduling Criteria Scheduling Algorithms.
What we will cover…  CPU Scheduling  Basic Concepts  Scheduling Criteria  Scheduling Algorithms  Evaluations 1-1 Lecture 4.
5: CPU-Scheduling1 Jerry Breecher OPERATING SYSTEMS SCHEDULING.
Political Economy of European Monetary Integration Europe in World Economy 2015 Vladan Hodulak.
mankiw's macroeconomics modules
Self-Organizing Agents for Grid Load Balancing Junwei Cao Fifth IEEE/ACM International Workshop on Grid Computing (GRID'04)
Lecture # 5 Role of Central Banks. Role of Central bank Monitoring Provide guide lines.
Chapter 5: CPU Scheduling. 5.2 Silberschatz, Galvin and Gagne ©2005 Operating System Concepts – 7 th Edition, Feb 2, 2005 Basic Concepts Maximum CPU utilization.
Copyright © 2006 by The McGraw-Hill Companies, Inc. All rights reserved. McGraw-Hill Technology Education Lecture 5 Operating Systems.
OPERATING SYSTEMS CPU SCHEDULING.  Introduction to CPU scheduling Introduction to CPU scheduling  Dispatcher Dispatcher  Terms used in CPU scheduling.
CHP-4 QUEUE.
Copyright © 2009 Pearson Addison-Wesley. All rights reserved. Chapter 21 Monetary Policy Strategy.
Frascati, October 9th, Accounting in DataGrid Initial Architecture Albert Werbrouck Frascati, October 9, 2001.
Frascati, October 5th, Accounting in DataGrid Preliminary Proposal and basis for discussion Stefano Barale Frascati, October.
Silberschatz, Galvin and Gagne  Operating System Concepts Chapter 6: CPU Scheduling Basic Concepts Scheduling Criteria Scheduling Algorithms.
1 520 Student Presentation GridSim – Grid Modeling and Simulation Toolkit.
Your university or experiment logo here Caitriana Nicholson University of Glasgow Dynamic Data Replication in LCG 2008.
Wolfgang EffelsbergUniversity of Mannheim1 Differentiated Services for the Internet Wolfgang Effelsberg University of Mannheim September 2001.
Silberschatz and Galvin  Operating System Concepts Module 5: CPU Scheduling Basic Concepts Scheduling Criteria Scheduling Algorithms Multiple-Processor.
Silberschatz, Galvin and Gagne  Operating System Concepts Chapter 6: CPU Scheduling Basic Concepts Scheduling Criteria Scheduling Algorithms.
Chapter 5: CPU Scheduling. 5.2 Silberschatz, Galvin and Gagne ©2005 Operating System Concepts Chapter 5: CPU Scheduling Basic Concepts Scheduling Criteria.
Chapter 5: Process Scheduling. 5.2 Silberschatz, Galvin and Gagne ©2005 Operating System Concepts Basic Concepts Maximum CPU utilization can be obtained.
1 11/29/2015 Chapter 6: CPU Scheduling l Basic Concepts l Scheduling Criteria l Scheduling Algorithms l Multiple-Processor Scheduling l Real-Time Scheduling.
RTOS task scheduling models
Authors: Rajkumar Buyya, David Abramson & Jonathan Giddy
Chapter 3 System Performance and Models Introduction A system is the part of the real world under study. Composed of a set of entities interacting.
6.1 CPU Scheduling Basic Concepts Scheduling Criteria Scheduling Algorithms Multiple-Processor Scheduling Real-Time Scheduling Algorithm Evaluation.
Memory Management OS Fazal Rehman Shamil. swapping Swapping concept comes in terms of process scheduling. Swapping is basically implemented by Medium.
1.  System Characteristics  Features of Real-Time Systems  Implementing Real-Time Operating Systems  Real-Time CPU Scheduling  An Example: VxWorks5.x.
Chapter 4 CPU Scheduling. 2 Basic Concepts Scheduling Criteria Scheduling Algorithms Multiple-Processor Scheduling Real-Time Scheduling Algorithm Evaluation.
Holding slide prior to starting show. Scheduling Parametric Jobs on the Grid Jonathan Giddy
Basic Concepts Maximum CPU utilization obtained with multiprogramming
Lecturer 5: Process Scheduling Process Scheduling  Criteria & Objectives Types of Scheduling  Long term  Medium term  Short term CPU Scheduling Algorithms.
1 Chapter 5: CPU Scheduling. 2 Basic Concepts Scheduling Criteria Scheduling Algorithms.
Enabling Grids for E-sciencE Claudio Cherubino INFN DGAS (Distributed Grid Accounting System)
Copyright ©: Nahrstedt, Angrave, Abdelzaher
Copyright ©: Nahrstedt, Angrave, Abdelzaher
Operating Systems Processes Scheduling.
COST PLANNING AND ESTIMATION
Chapter 6: CPU Scheduling
CPU Scheduling G.Anuradha
Module 5: CPU Scheduling
Operating System Concepts
3: CPU Scheduling Basic Concepts Scheduling Criteria
Chapter5: CPU Scheduling
Authors: Oren Rigbi Damian Goren
Chapter 6: CPU Scheduling
CPU SCHEDULING.
Chapter 5: CPU Scheduling
Wide Area Workload Management Work Package DATAGRID project
Uniprocessor scheduling
Operating System , Fall 2000 EA101 W 9:00-10:00 F 9:00-11:00
Chapter 6: CPU Scheduling
A pre-alpha accounting architecture
Module 5: CPU Scheduling
Chapter 6: CPU Scheduling
Module 5: CPU Scheduling
Presentation transcript:

- DataGrid meeting in Catania – 9 th to 11 th of April, 2001 A pre-alpha accounting architecture The “Computational Economy” model applied to DataGrid accounting

- DataGrid meeting in Catania – 9 th to 11 th of April, 2001 The accounting problem in brief Services required from the accounting architecture: Optimization of queues load. Determining the priorities for accessing execution queues on the resources (allocation policies). Meeting of job requirements (cpu speed, storage space, deadlines for receiving the output back). Obviously this accounting architecture will have to cope with the heterogeneity of the resources and of the local accessing policies.

- DataGrid meeting in Catania – 9 th to 11 th of April, 2001 What is Computational Economy? In recent years the problem of accounting has been investigated by many theoretical groups giving birth to various theoretical models and algorithms. The most flexible and capable of meeting our needs seems to be an economical analogy model. In this picture resources (i.e. cpu time) are exchanged between the groups participating in the project using a kind of virtual money (that we'll call Grid Credits, GC). Each time a group allows the use of its resources gains some GC that can be spent to get access to other Grid resources when needed.

- DataGrid meeting in Catania – 9 th to 11 th of April, 2001 A theoretical model What distinguish among the various theoretical models (and the extent of the economical analogy) is the price-setting mechanism used to decide the price of resource access and the type of algorithm used to reach market equilibrium. This last goal is of prime importance to avoid the trigger of undesirable effects like ‘virtual’ inflation or ‘virtual’ finance speculation.

- DataGrid meeting in Catania – 9 th to 11 th of April, 2001 Price-setting policies Here's a list of possible price-setting policies we found in literature: 1. Commodity market (flat or demand & supply driven pricing) model 2. Posted-price model 3. Bargaining model 4. Tendering/contract net model 5. Auction Model 6. Bid-based proportional resource sharing model 7. Community/coalition/bartering model

- DataGrid meeting in Catania – 9 th to 11 th of April, 2001 Price-Oriented and Resource-Oriented Algorithms (1) As we noticed before, the possibility of bulding a satisfying model heavily relies on the certainty of avoiding the typical undesireable effects of real world markets. This goal is within reach if the system is constrained to reach and keep equilibrium (while real world markets usually expand). In literature this kind of model is usually called 'multy-commodity market' (where each Grid site is the single commodity). The search for general equilibrium in such a system has been performed with basically two families of algorithms:

- DataGrid meeting in Catania – 9 th to 11 th of April, 2001 Price-Oriented and Resource-Oriented Algorithms (2) 1. Price-Oriented Algorithms. This algorithms guarantee that the price of each commodity is equal for every group participating in the Grid, but some groups may be unable to access Grid resources. 2. Resource-Oriented Algorithms. This ones are based on the resource constraint which guarantees that allocation is always feasible, since resources are only transferred between groups participating in the Grid.

- DataGrid meeting in Catania – 9 th to 11 th of April, 2001 Basic Principles In our preliminary work we started from the following basic principles: 1. A Grid structure with FIFO, FCFS or other 'simple' priority schemes won't fit the needs. 2. No one would participate in the Grid if there are no advantages in doing it (the flop of experiences like Popular Power seems to demonstrate this assumption quite clearly).

- DataGrid meeting in Catania – 9 th to 11 th of April, 2001 Proposals for guidelines (1) 1. Establish a biunique correspondance between Grid Credits and the amount of resources shared (no one is allowed to coin GC). This forces the model into a much simpler pure exchange and resource oriented system. 2. The virtual account which manages each user's GC is located at her/his Home Location Registry (HLR, in analogy with the one used by mobile phones networks). 3. Manage GC exchange with debit card like mechanism. This guarantees no-one could easily forge GC and lets taking into account GC that are gained by the user while using 'foreign' Grid resources.

- DataGrid meeting in Catania – 9 th to 11 th of April, 2001 Proposals for guidelines (2) 4. At each Grid resource, the cost for a job submitted to a particular queue is expressed in some simple way (regardless of the complexity of the issue of calculating it) 5. Since we want to assign cost to resources in an homogenous way, each time a new resource enters the Grid a benchmarking utility is automatically run and ends sending (in a sealed way) the benchmarking results to some appropriate Information Service. To help the end user in estimating realistic computing costs this benchmarks can also include typical applications (maybe like the reconstruction of a HEP event?).

- DataGrid meeting in Catania – 9 th to 11 th of April, 2001 A possible accounting flow (1) A job request arrives to the queue manager that responds to the submitter giving: a) lower cost queue where the job should be executed within the time limit specified by the submitter; b) cost of execution assuming estimated time should be associated with a tolerance or overrun, such as “2 hrs + 1 hr tolerance”. User replies confirming or canceling reservation within, say, ten minutes. If no reply arrives after above time, job is definitely inserted in queue and data transfer requested to WP2, if necessary. The HLR is checked to control that account balance is sufficient to cover the computing cost (estimated + tolerance).

- DataGrid meeting in Catania – 9 th to 11 th of April, 2001 A possible accounting flow (3) The amount of GC corresponding to the maximum duration (estimated + tolerance) is 'frozen' before the job is finally submitted. When the job finishes the frozen amount is released and the real cost is deducted from GC balance. GC which come into computing facilities associated with a HLR will be distribuited according to local criteria, not part of the Grid middleware.

- DataGrid meeting in Catania – 9 th to 11 th of April, 2001 References R. Buyya, D. Abramson, J. Giddy - “A Case for Economy Grid Architecture for Service Oriented Grid Computing” F. Ygge / H. Akkermans “Duality in Multi-Commodity Market Computations” - An example of a ‘Grid like’ environment with price setting policy based on a community/coalition/bartering model The first ‘Grid flop’ we know of: