grid & virtualization working group

© 2007 Open Grid Forum

Grid & Virtualization Working Group

OGF20 gridvirt-wg

May 7, 2007Manchester, UK

Erol Bozak, ChairSAP, Development Architect

Wolfgang Reichert, Co-ChairIBM, Senior Technical Staff Member

2© 2007 Open Grid Forum

• “I acknowledge that participation in this meeting is subject to the OGF Intellectual Property Policy.”• Intellectual Property Notices Note Well: All statements related to the activities of the OGF and addressed to

the OGF are subject to all provisions of Appendix B of GFD-C.1, which grants to the OGF and its participants certain licenses and rights in such statements. Such statements include verbal statements in OGF meetings, as well as written and electronic communications made at any time or place, which are addressed to:

• the OGF plenary session, • any OGF working group or portion thereof, • the OGF Board of Directors, the GFSG, or any member thereof on behalf of the OGF, • the ADCOM, or any member thereof on behalf of the ADCOM, • any OGF mailing list, including any group list, or any other list functioning under OGF auspices, • the OGF Editor or the document authoring and review process

• Statements made outside of a OGF meeting, mailing list or other function, that are clearly not intended to be input to an OGF activity, group or function, are not subject to these provisions.

• Excerpt from Appendix B of GFD-C.1: ”Where the OGF knows of rights, or claimed rights, the OGF secretariat shall attempt to obtain from the claimant of such rights, a written assurance that upon approval by the GFSG of the relevant OGF document(s), any party will be able to obtain the right to implement, use and distribute the technology or works when implementing, using or distributing technology based upon the specific specification(s) under openly specified, reasonable, non-discriminatory terms. The working group or research group proposing the use of the technology with respect to which the proprietary rights are claimed may assist the OGF secretariat in this effort. The results of this procedure shall not affect advancement of document, except that the GFSG may defer approval where a delay may facilitate the obtaining of such assurances. The results will, however, be recorded by the OGF Secretariat, and made available. The GFSG may also direct that a summary of the results be included in any GFD published containing the specification.”

• OGF Intellectual Property Policies are adapted from the IETF Intellectual Property Policies that support the Internet Standards Process.

OGF IPR Policies Apply


• Status of GridVirt Activities

• Management of Virtual Machines in Grid Infrastructures• Speaker: Ruben S. Montero

• Use Cases

• Virtual Workspaces

• Work Streams for the GridVirt-Working Group

Agenda


Status

• Introduction of the working group at OGF19• Virtualization Concepts• Goals and Scope Definition• Milestones Definition

• Milestones• Milestone 1 (OGF 19)

• Introduction, scope definition & milestones definition• Milestone 2 (OGF 20)

• Terminology definition • Collection of use cases• Determine relations to other OGF WGs and SDOs

• Milestone 3 (OGF 21)• Requirements collection• Determine relation to other standards• First draft of a profile


EGA Reference Model

Grid Management Entity

Grid Component

Policies

Monitor

Enterprise

Other GME

Accounting / Billing

Manage

External Events

The role of Grid Management Entity (GME)


GCGC

Logical GME Grid Components & Dependencies

EGA Reference Model


Grid Component

GC …

…

Monitor

GC GCOther GME

Manage

Policies

Enterprise


Policies Accounting / Billing

Manage

Monitor

GCGCGC


Logical GME

Reference Model – Virtualization

VS Scenario /Landscape

Virtualization Platform GME

VS System GME

Application GME Application

Hypervisor

VSVSVS

Manage

Monitor

Monitor

Manage

Policies Accounting / Billing

PoliciesAccounting / Billing

Grid Components & Dependencies

Manage

Monitor


EGA Reference Model


Grid Component

Policies

Monitor

Enterprise


ManageResources (other GCs)

Resources (other GCs) MetricsMetrics

Service Level Management

Assigns / Provisions

Reconciles

ConsumedGenerates


EGA Reference Model

Lifecycle of a Grid Component

Unconfigured

Inactive

Active

Create / Discover Destroy

Configure Unconfigure

Start Stop

Provision Decommission

Manage


Use Cases Structure

System Virtualization

Provision Virtual System(s)


Manage Virtual System(s)


Decommission Virtual System(s)


Create / Discover Image(s)

Create / Discover Image(s)

Configure Image(s)

Configure Image(s)

Deploy Virtual System(s) from

Image(s)

Deploy Virtual System(s) from

Image(s)


Use Cases Structure

MigrationMigration

Dynamic Resizing

Dynamic Resizing

MonitoringMonitoring

SnapshottingSnapshotting

System Virtualization








Virtualization Use Cases

Migration of virtual system during runtime• Power Saving

The resource manager may pool virtual systems to a reduced number of physical systems in order to save power.

• Planned maintenanceThe physical system as well as the hypervisor could require maintenance activities to be performed (e.g. install patch, hw upgrade or driver etc.). The running job could be migrated to other machines without downtime.

• Adaptation to changing capacity requirements & conditions (availability or offering)

Capacity availability may change in the environment because of recently freed resources by the completed jobs or additional physical systems may be introduced.



Power Saving• Policy-driven• Monitor event

• Resource utilization below threshold

• Temperature above threshold

• External event • From hierarchically higher GME

Resulting management actions• Communication with System

Virtualization GME(s) for live migration• Resource allocation / deallocation• Notification of grid component (before

and after live migration)• Accounting event


Grid Component

Policies

Monitor

Enterprise





Reconciles

ConsumedGenerates



Planned Maintenance• External event


Virtualization GME(s) for live migration• Resource allocation / deallocation• Notification of grid component (before

and after live migration)• Accounting event

GMEGrid

Component

Policies

Monitor

Enterprise

Other GME


Manage

External Events



Dynamic Resizing• Dynamically changing capacity requirements

During runtime the job may require additional capacity (e.g. CPU capacity, Memory capacity, I/O bandwidth etc.). If the underlying physical system is able to serve the requirements more capacity for the job / virtual system can be provided locally on the same physical system.

• Dynamically changing capacity offering / availabilityCapacity availability may change in the physical system (e.g. CPU capacity, Memory capacity, I/O bandwidth etc.) because of recently freed resources by the completed jobs. In these situations available capacity can be utilized for the running jobs.



Dynamic Resizing• Policy-driven• Monitor event

• SLO is going to be missed

(progress indicator, trend analysis)

• Grid component event• Dynamic resource requirements

• External event • Dynamic resource availabiliy


Virtualization GME for system resizing• Virtual resource reallocation• Changing system parameters/settings• Notification of grid component (before

and after resizing)• Accounting event


Grid Component

Policies

Monitor

Enterprise





Reconciles

ConsumedGenerates



Snapshotting• Stateful cloning

The execution of a job may require costly preparation steps, e.g. retrieving data from the backend, which might be common to all jobs of an activity / application. Rather than doing the preparation separately for each job one (or the first) job can be snapshotted after the preparation and the state can be cloned and distributed.

• Reproducing situationsFor purpose of diagnosis (error or performance analysis) the user may repetitively re-run the same job from a certain persisted state.

• Protecting (long running) jobs from software or hardware failures

By providing recovery points that can be re-activated (long running) jobs can be restarted at a certain persisted state potentially on a different physical system.



Stateful Cloning• Policy-driven or external request

from top-level GME

• Subsequent provisioning scenario using the snapshot


Virtualization GME to take snapshot• Notification when snapshot has been

taken• Accounting event

Top-level GME manages• Cloning of the snapshot (distribution,

postprocessing)• Provisioning scenario

GMEGrid

Component

Policies

Monitor

Enterprise

Other GME


Manage

External Events



Reproducing Situations• External event to take snapshot

• External event to restart from persisted state


Virtualization GME to take snapshot

• Communication with System Virtualization GME to restart from snapshot

GMEGrid

Component

Policies

Monitor

Enterprise

Other GME


Manage

External Events



Isolation• Metering of job resource consumption

For the purpose of accounting and billing.

• Resource consumption controlThrough isolation the amount / degree of resource consumption can be controlled and leveled, i.e. greedy jobs can be controlled.



Metering of Job Resource Consumption

• Data collection at deprovisioning event


Virtualization GME to get accurate metering data for the lifetime of the virtual system

GMEGrid

Component

Policies

Monitor

Enterprise

Other GME


Manage

External Events



Resource Consumption Control• Policy-driven


Virtualization GME to set limits

GMEGrid

Component

Policies

Monitor

Enterprise

Other GME


Manage

External Events



Provisioning Scenarios• Resource Provisioning

Definition and activation of the desired runtime environment of a job: Rather than searching for and allocating resources for the job resources can be “created” on demand.

Definition and provisioning of the required software stack (runtime environment)

• EmulationEmulating an environment for legacy jobs: Legacy applications / jobs may require certain physical resources or a certain runtime environment (e.g. operating system). In this situation a virtual system may emulate the legacy environment.

• IsolationAvoiding conflicts.

Ensuring security: To protect the job from spyware the job can be executed in its own dedicated and certified virtual system.


Virtual Workspaces

• A Virtual Workspace is an abstraction of an execution environment…• …that can be made dynamically available • …through well defined protocols,• …the software environment contained in the workspace and

the user submitting the workspace are both trustworthy.

• Virtual Workspaces is not a new idea!• Dynamically setting up cluster nodes

• CoD: http://www.cs.duke.edu/nicl/cod/• bcfg: http://trac.mcs.anl.gov/projects/bcfg2/

• • Providing access to existing installation• Dynamic Accounts: http://workspace.globus.org/da/

• Refining site configuration• Pacman: http://www.archlinux.org/pacman/


Virtual Workspaces

Representation of a Virtual Workspace

Virtual Workspace SpecificationVirtual Workspace

Deployment

XMLXML

MetadataVM Image

XMLXML

Deployment Request


Virtual Workspaces

• Specification of a Virtual Workspace• VM Image• Metadata

• XML Document• Includes deployment-independent information:

VMM and kernel requirementsNICs + IP configuratoinVM image location

Don‘t changes between deployments

• Deployment Request• Specifies availability, memory, CPU%, disk Changes during or between deployments


VW Node

Virtual Workspaces

Node Pool

Node

Node

Node

Node

Node

Node

Node

Node

Node

Node

Node

Node

VW Factory Service

Image Node

VW Service

Trusted Computing Base


VW Node

Virtual Workspaces

Node Pool

Node

Node

Node

Node

Node

Node

Node

Node

Node

Node

NodeVW Factory Service

Image Node

VW Service


Manage- Start / Stop / Suspend- Migrate- Monitor etc.

Create- Metadata Instance- Deployment Req.

Node

Agent Hypervisor

Node


VW Node

Virtual Workspaces

Node Pool

Node

Node

Node

Node

Node

Node

Node

Node

Node

Node

Node

Node

VW Factory Service

Image Node

VW Service




VW Node

Virtual Workspaces

Node Pool

Node

Node

Node

Node

Node

Node

Node

Node

Node

Node

Node

Node

VW Factory Service

Image Node

VW Service


Manage- Start / Stop / Suspend- Migrate- Monitor etc.


Workstreams

• Workstream 1: Refine Use Cases & align Grid Reference Architecture in the Context of System Virtualization• Define the requirements to the grid architecture for

integration with system virtualization platforms

• Workstream 2: Refine the Provisioning Use Case• Define Interaction among the components in the architecture

to create / discover, configure and start a Virtual System• Define information model for definition of Virtual Systems Exploit the concept of „Virtual Workspaces“


Appendix

• Project Homepage• https://forge.gridforum.org/sf/projects/gridvirt-wg

• Mailing list• [email protected]• Subscription: http://www.ogf.org/mailman/listinfo/gridvirt-wg

grid & virtualization working group

Documents