We think you have liked this presentation. If you wish to download it, please recommend it to your friends in any social system. Share buttons are a little bit lower. Thank you!
Presentation is loading. Please wait.
Published byClaire Payne
Modified over 3 years ago
© 2006 Open Grid Forum Advanced reservation/Scheduling Tomohiro Kudoh (AIST, Japan)
© 2006 Open Grid Forum 2 Types of request Designate request A request of specific physical resources Abstract request A request of resources with specified capabilities (attributes). Mapping to physical resources required.
© 2006 Open Grid Forum 3 Scheduling and provisioning Scheduling Allocating using right of resources to a request If the request is an abstract request, mapping between the request and physical resources is also determined Provisioning Activate allocated resources and make them usable by the requester. May be automatic / may be kicked by additional operation by the requester (signaling)
© 2006 Open Grid Forum 4 Types of scheduling (1) On-demand If the resources are available when a request is issued, the resources are allocated to the requester. If the request is an abstract request, mapping of physical resources is done. If the resources are not available, the request is simply denied. Usually, the requester can use the allocated resources as long as it wants. The end time of the provisioning is not determined at the time of provisioning.
© 2006 Open Grid Forum 5 Types of scheduling (2) Batch Widely used in computing systems such as supercomputing centers. Requests are queued in a queue, and the request at the head of the queue is scheduled when required resources become available If the request is an abstract request, mapping of physical resources is done at this moment. May have multiple prioritized queues.
© 2006 Open Grid Forum 6 Types of scheduling (3) Advance reservation (book ahead) A request is processed by a scheduler, and the scheduler finds a period when the requested resources are available for the requested duration The resources are reserved during the reservation period, and when the reservation period begins, the resources are allocated to the requester. A reservation database is maintained by the scheduler or resource managers, and referred and updated by the scheduler. Scheduling is done when a request is issued. In addition, re- scheduling may be done when availability of resources is changed.
© 2006 Open Grid Forum 7 Types of scheduling (3) Instant reservation If an advance reservation request requests immediate allocation of resources, the request is called instant reservation. Different from the pure on-demand the end time of a provisioning is determined at the time of scheduling.
© 2006 Open Grid Forum 8 Priority and preemption If requests have priorities, a request with lower priority may be preempted by a higher priority request Provision preemption To preempt resources which are being provisioned (and may be being used by the requester) to a lower priority request. Schedule preemption To preempt an already scheduled but not yet provisioned preceding lower priority request
© 2006 Open Grid Forum 9 Priority and preemption If provision preemption is supported, both on-demand and advance reservation requests can be used simultaneously. When the same resources are required, advance reservation requests with higher priority than existing (being provisioned) on-demand requests can be scheduled At the beginning of the reserved period of the advance reservation request, the resource is preempted (if the resource is still being used by the on-demand request) If schedule preemption is supported in an advance reservation system, re-scheduling may be done. After the re-scheduling, reserved period or other conditions of existing lower priority requests may be changed. A mechanism to let the clients of such existing requests know the change may be required.
© 2006 Open Grid Forum Summary 10 Timing of Scheduling (to request) Timing of Provisioning (to scheduling) End time of provisioning Maintained states (DB) Schedule preemption Provision preemption On-demandinstantaneous Not determined in advance Current state of resources Not applicable May be supported, required for co-use with reservation Instant reservation instantaneous Determined at the time of scheduling Current state of resources, Reservation database Not applicable May be supported Advance reservation Instantaneous/ when availability of resources changed scheduledDetermined at the time of scheduling Current state of resources, Reservation database May be supported BatchWhen a request comes to the top of the queue instantaneousNot determined in advance Current state of resources, Queue(s) of requests May be supported May be supported, required for co-use with reservation
© 2010 Open Grid Forum Choices in Reservation operation 1PC/2PC, timing Tomohiro Kudoh, AIST.
National Institute of Advanced Industrial Science and Technology Advance Reservation-based Grid Co-allocation System Atsuko Takefusa, Hidemoto Nakada,
TeraGrid Advanced Scheduling Tools Warren Smith Texas Advanced Computing Center wsmith at tacc.utexas.edu.
CS 149: Operating Systems February 3 Class Meeting
© 2004, D. J. Foreman 1 Scheduling & Dispatching.
Service Differentiation in Optical Burst Switched Networks Chi-Hong Loi, Wanjiun Liao, and De-Nian Yang GLOBECOM’02.
Peter Couvares Associate Researcher, Condor Team Computer Sciences Department University of Wisconsin-Madison
© 2006 Open Grid Forum Network Service Interface in a Nut Shell GEC 19, Atlanta, GA Presenter: Chin Guok (ESnet) Contributors: Tomohiro Kudoh (AIST), John.
© 2006 Cisco Systems, Inc. All rights reserved. MPLS v2.2—8-1 MPLS TE Overview Understanding MPLS TE Components.
Presented by: Priti Lohani
Scheduling in HPC Resource Management System: Queuing vs. Planning Matthias Hovestadt, Odej Kao, Alex Keller, and Achim Streit 2003 Job Scheduling Strategies.
Resource Management Reading: “A Resource Management Architecture for Metacomputing Systems”
© 2006 Cisco Systems, Inc. All rights reserved.QoS v2.2—5-1 Congestion Management Configuring LAN Congestion Management.
Lottery Method in BPA’s Simultaneous Submission Windows and Potential Application to the Simultaneous Defender Matching Process NAESB Meeting 05/01/13.
Lucy Yong Young Lee IETF CCAMP WG GMPLS Extension for Reservation and Time based Bandwidth Service.
1 Network Services Interface An Interface for Requesting Dynamic Inter- datacenter Networks Tomohiro Kudoh (AIST) Guy Roberts (DANTE) Inder Monga (ESnet)
The Glidein Service Gideon Juve What are glideins? A technique for creating temporary, user- controlled Condor pools using resources from.
Short-Term Competitions and Preemption. Overview Standards being covered are: Motion 2 - Fixed Capacity Over Term of Request. ▪ Tier 1 Service –
Scientific Workflow Scheduling in Computational Grids Report: Wei-Cheng Lee 8th Grid Computing Conference IEEE 2007 – Planning, Reservation,
Introduction to Embedded Systems Resource Management - III Lecture 19.
© 2017 SlidePlayer.com Inc. All rights reserved.