enterprise grid alliance - open grid forum use cas… · enterprise grid alliance accelerating the...

Enterprise Grid AllianceAccelerating the Deployment of Grids in Enterprise Data Centers

Paul StrongSystems Architect, Sun Microsystems, Inc.Chair, EGA TSC & Ref Model WG

GGF15 - Capturing Enterprise GridRequirements Through Use Cases

12/2/2005GGF14 - OGSA/EGA Session Copyright 2005, Enterprise Grid Alliance 2

Enterprise Grid Computing• Different priorities to traditional Grid community

– All about guarantees and not just best efforts– Mission critical applications

• Commercial and technical workload, inc. ALL of– Traditional mixed workload, transactional (OLTP) and batch– Data warehousing– Web services and service oriented architectures– Compute and I/O intensive

• Key benefits - economies of scale– Better performance, scaling, throughput and resilience– Greater efficiency and agility

• Inhibitors To Adoption– Grid is poorly understood– MUST NOT break today’s processes


Enterprise Grid Alliance Working Groups• Reference Model

– Common lexicon/taxonomy/model– Use cases

• Component Provisioning– Requirements for provisioning (including use cases)– Initially focused on servers

• Data Provisioning– Provisioning and management of data

• Security– Requirements exclusive to grids

• Utility Accounting– Requirements for enabling utility billing and pricing


EGA Reference Model: Glossary• An enterprise Grid is a collection of interconnected

(networked) Grid components under the control of a Grid Management Entity (GME)

• Grid component - A super class of object including everything that is managed in an enterprise Grid– Physical: servers, disks, switches, etc.– Logical: services, applications, operating systems, load balancing

software, tiers of services, etc.

• Grid management entity - the logical aggregation that managed the enterprise Grid– People– Process– Technology


EGA Reference Model: Representing Services As Graphs

OS InstanceOS Instance

ServerServer


HardwarePartition

HardwarePartition

ServerServer

DatabaseInstance

DatabaseInstance

DatabaseInstance

DatabaseInstance

ApplicationServer

Instance

ApplicationServer

Instance

WebServer

Instance

WebServer

Instance

BusinessLogicTier

BusinessLogicTier

PresentationTier

PresentationTier

E BookstoreE Bookstore

PersistentStorage

Tier

DatabaseCluster

ApplicationServerCluster

WebServerFarm

LoadBalancer��

LoadBalancer��

App ServerLoad

Balancer��

App ServerLoad

Balancer��

Firewall��Firewall�� LoadBalancer��Load

Balancer�� FirewallFirewall

AppServer

Firewall��

AppServer

Firewall��

WebServerFirewall

WebServerFirewall

Web ServerLoad

Balancer��

Web ServerLoad

Balancer��

ServerServer

OS InstanceOS Instance OS InstanceOS Instance

ServerServerBoot

Disk/LUNBoot

Disk/LUNOS MediaOS Media

DatabaseStorage

DatabaseStorage

More detailed drill down

PatchesPatchesDatabaseMediaDatabase

Media

OS ContainerOS ContainerDatabase

TransactionLog Storage

DatabaseTransactionLog Storage


PhysicalPhysical

Virtualized Physical

OE

Virtualized OE

Platform Instance

Virtualized Platform

Biz Process/-Service

Storage NetworkComputeCompute

Aggregations

Web Server Farm Federation

Clusters Load Balanced Farms

EE--BookstoreBookstore ERP ServiceERP Service Online BankOnline Bank

DatabaseDatabase

NetworkFile systems -

NFS, CIFS

LDAPLDAP

Web ServerWeb Server Application Application ServerServer

File SystemsFile Systems

LUNs,Volumes

Disks, Array Controllers, SAN Switches

Virtualized OSeg N1 Grid Containers, BSD

Jails etc.

OS OS -- eg AIX, HP/UX, eg AIX, HP/UX, Linux, Solaris, Linux, Solaris, Windows etc.Windows etc.

VMMs & Hypervisors

Hardware Partitions

Servers,Servers,Blades etc.Blades etc.

Load Balancers, Global IP in

clusters

IP, TCP, UDP etcIP, TCP, UDP etc

VLANs

Switches, Routers etc..

EGA Reference Model: Classifying Grid Components


EGA Reference Model: Grid Component Life Cycle

ProvisionProvisionProvision DecommissionDecommissionDecommission

ManageManageManage


EGA Reference Model: Representing Services As Graphs


ServerServer


HardwarePartition

HardwarePartition

ServerServer

DatabaseInstance

DatabaseInstance

DatabaseInstance

DatabaseInstance

ApplicationServer

Instance

ApplicationServer

Instance

WebServer

Instance

WebServer

Instance

BusinessLogicTier

BusinessLogicTier

PresentationTier

PresentationTier

E BookstoreE Bookstore

PersistentStorage

Tier

DatabaseCluster

ApplicationServerCluster

WebServerFarm

LoadBalancer��

LoadBalancer��

App ServerLoad

Balancer��

App ServerLoad

Balancer��

Firewall��Firewall�� LoadBalancer��Load

Balancer�� FirewallFirewall

AppServer

Firewall��

AppServer

Firewall��

WebServerFirewall

WebServerFirewall

Web ServerLoad

Balancer��

Web ServerLoad

Balancer��

ServerServer

OS InstanceOS Instance OS InstanceOS Instance

ServerServerBoot

Disk/LUNBoot

Disk/LUNOS MediaOS Media

DatabaseStorage

DatabaseStorage

More detailed drill down

PatchesPatchesDatabaseMediaDatabase

Media





Scenarios & Use Cases• Scenario

– The storyboard, e.g. line of biz VP walks into IT VP’s office and says “I need this new service deployed”

• Use Case– Enumerated set of steps required to realize the scenario– Sequence diagrams (UML)– Detailed– Leverage EGA reference model– Describes today’s process (NOT the desired state) so as to identify

major pain points and their context– Goal

• Automate process by replacing actors with software/services• Identifies appropriate services and their functionality


Use Cases• Key Enterprise Grid Technology Areas

– Virtualization– Abstraction– Automation

• Automation– Moving Process From Human To Machine Domain

• Enterprises Are Built On Process– ITIL, eTOM, COBIT, Ad Hoc

• Use Cases– Proven mechanism to aid problem decomposition– Allow us to enumerate and capture problem and context, including

process– Leverage Reference Model - map processes to model


EGA Reference Model: Use Case -Managing Grid Component Life Cycle


Use Case Format• Title• Summary• Dependencies• Actors• Stakeholders• Pre-Conditions• Initiation Of Use Case• Flows

– Basic, Alternate, Exceptional

• Success Criteria• Post Conditions


Provisioning By Walking The DAG


ServerServer

DatabaseInstance

DatabaseInstance

BootDisk/LUN

BootDisk/LUNOS MediaOS Media

DatabaseStorage

DatabaseStorage

OSPatches

OSPatches




DatabaseMedia

DatabaseMedia

DatabaseBinary Storage


ApplyVerb {ProvisionGridComponent}


Provisioning By Walking The DAG


ServerServer

DatabaseInstance

DatabaseInstance

Boot Disk/LUNBoot Disk/LUNOS MediaOS Media

DatabaseStorage

DatabaseStorage

OSPatches

OSPatches




DatabaseMedia

DatabaseMedia



From left to right

ApplyVerb {ProvisionGridComponent}

From

leaf

nod

es u

p1 2

3.1.1 3.1.2

3.1

3.1.3

3.2

3 4 5


EGA Use Case Delivery• Pragmatic

– Driven by high priority data center problems– Context for other EGA working groups– Basis for collaboration with other bodies, e.g. GGF, DMTF, SNIA

• v1.0 - Released May 2005– EGA Reference Model– Template use case

• v Next - Release Oct/Nov 2005– Focus on provisioning– Bottom up– 70+ detailed use cases– Validation via User Forums/Group

• v Next++– Service Level Management use cases– Top down


RequirementsInhibitors To Adoption

EGAEGAEGA

EnterpriseUsers*

EnterpriseEnterpriseUsers*Users* Education

Reference ModelProfilesStandardsReference ArchitecturesProof PointsDemonstrations

Standard Interfaces/APIsArchitectures

CompliantImplementations,Products

StandardsBodies

StandardsStandardsBodiesBodies

Requirements, Model,Glossary, Use Cases

VendorsVendorsVendors *Enterprise users•Commercial end users•Public sector end users•Systems integrators•Solutions providers

*Enterprise users•Commercial end users•Public sector end users•Systems integrators•Solutions providers

Requirements, Model,Glossary, Use CasesStandards

EGA Role as Requirements Definer


General Information:

[email protected]

Sign Up to Stay Informed. Join the EGA Interest List:

www.gridalliance.org/interest/

For More Information


Decommission Grid Component Use Case


Monitor & Manage Grid Component Use Case


Grid Computing Landscape• Network scale computing leverages

– Trend toward network distributed services– Network bandwidth

• Potentially unbounded– Performance + scaling + throughput– Inherent resilience

• Widespread adoption, but still largely perceived as…– Academic and scientific workload– Compute intensive

• Growing ecosystem– Product and service providers– Standards setting organizations


Barriers to Enterprise Grid Adoption• Reluctance to risk mission critical applications• Confusion

– What is Grid?– How to apply grid in commercial data centers?

• Immaturity– Incomplete models– Lack of stable standards– Shortage of case studies and proven roadmaps– Unproven interoperability between vendor products

• Additional inhibitors– Cultural attitudes toward exclusive ownership of IT assets– Trust and accounting model issues– Licensing and resource sharing concerns


Enterprise Grid Alliance• Vendor neutral consortium

– 30+ leading vendors and end-users

• Global participation– HQ in US, regional steering committees in Japan and EMEA

• Clear mission– Accelerating the deployment of Grids in enterprise data centers– Single focus: enterprise data centers

• Not a traditional standards body– Achieve results through collaboration– Create standards as a last resort

• Pay-to-play level playing field– Open to ALL stakeholders– One organization one vote


EGA Membership Growth – 32(60 percent growth since April 2004 launch )


Enterprise Grid Alliance Reference Model• Fills an obvious gap

– complementary to the existing body of work

• Technology agnostic– Makes no assumptions about implementation– Avoids need to re-write/re-factor as technologies evolve

• Vendor neutral– 20 participants from 14 EGA member organizations

• Catalyst– Reflection of the current understanding of Grid stakeholders– Basis for collaboration with other industry bodies to ensure emerging

standards meet enterprise requirements

• Evolution– Validate the model – Solicit feedback From GGF, DMTF,SNIA, vendors, SI’s and end users


EGA Technical Program• Simple strategy

– Educate– Capture requirements– Drive interoperability– Drive adoption

• Pragmatic approach– Non-traditional; delivery through collaboration– Consensus driven application of domain expertise– Focus on short to medium term deliverables– Phased approach to solutions

• Start simple - single data center with enterprise applications• Extend - include multiple data centers and technical applications• Unify and complete - bring together to reflect Grid as a utility

– Individually chartered working groups to address specific areas• Reference Model, Component Provisioning, Data Provisioning, Security And Utility

Accounting


EGA Reference Model• Glossary

– Defines a framework for classifying Grid resources/services together with their relationships and dependencies in a conceptual component architectural setting

• Model– Context for requirements, solutions and comparisons– Vendor neutral– Technology/implementation agnostic– Describes existing data centers– Does not assume a particular technology roadmap

• Use cases– Set of commercial enterprise community-centric use cases – Consistent and relevant requirements for partner SDOs and all

other enterprise Grid stakeholders


EGA Reference Model• Specific

– Enterprise data center grids

• Sharp focus– Provide shared context or framework for describing enterprise Grid

customer requirements, technical problems and technical solutions inside and outside EGA

• Simple– Small set of simple objects, relationships and verbs describe highly

complex data centers, their constituent parts and their life cycles

• Complementary– Addressing a gap


General RequirementsInhibitors To Adoption

EducationReference ModelStandardsProfilesReference ArchitecturesProof PointsDemonstrations

Standard Interfaces/APIsArchitectures

ValidationCompliantImplementations,Products

StandardsBodies

StandardsStandardsBodiesBodies

Requirements, Model,Glossary, Use Cases

VendorsVendorsVendors

EnterpriseUsers

EnterpriseEnterpriseUsersUsers

Specific Requirements(eg Security, Component Provisioning etc.)Specific Use Cases

EGAWG

EGAEGAWGWG Specific RequirementsEGA

ReferenceModel

EGAEGAReferenceReference

ModelModel

Requirements, Model,Glossary, Use Cases,Standards

Role of EGA Reference Model


Enterprise Grid - Definition• An Enterprise Grid is a collection of interconnected

(networked) Grid Components under the control of a Grid Management Entity.

• An Enterprise Grid is typically differentiated from traditional data centers by management practices and technology which– Make management service or application centric rather than

component centric – Enable the pooling and sharing of networked resources

• Initial scope for EGA working groups– Within a single data center– Under the control of a single entity– Commercial workloads


Datacenter Customers

Datacenter Component Manufacturers

Grid Resources

Enterprise Enterprise ApplicationsApplications(Services)(Services)

Grid Management EntityGrid Management EntityGrid Management Entity view

viewvi

ew

Datacenter O

perators

An Enterprise Grid


The Grid Component• Grid Component

– Is a super class of object covering every managed thing in the data center• Resources - e.g. servers, switches, disks, arrays, routers• Applications and Services - databases, ERP services, CRM etc.• Everything in between, e.g Meta-Resources such as

– Virtual Machines– Hardware Partitions– OS Partitions, such as BSD Jails etc.– Clusters

– Is recursive• A component such as an ERP service is clearly comprised of other components

such as databases, application servers etc. • These are in turn realized through binding to other components, i.e. more

traditional resources such as servers and operating systems etc.

– Has a life cycle• It is Created, Discovered, Provisioned, Managed, De-Commissioned, Destroyed


Service vs Resource• Whether a grid component is a service or a resource

is really a matter of perspective• The EGA ref model chooses not to make a distinction

- they are all managed objects


Grid Grid Grid

AbstractApplicationDefinition

AbstractTo ConcreteMapping

Data CenterData CenterDevices Devices --Elemental GridElemental GridComponentsComponents

FirewallPool

LoadBalancer

Pool

NetworkDevicePool

ServerPool

StoragePool

Business Service(for example eBanking app of Bookstore etc.)

Business Service(for example eBanking app of Bookstore etc.)

Mapping Business Services ToGrid Components


BookstoreAbstractApplicationDefinition


Network StorageFirewallsLoad

Balancers Servers

WebServer

Tier

WebServer

Tier

Database/PersistentStorage

Tier

Database/PersistentStorage

Tier

ApplicationServer

Tier

ApplicationServer

Tier



Grid Grid Grid


BookstorePresentationTier

ApplicationLogicTier

AbstractApplicationDefinition


PersistentStorage

Tier

Web ServerLoad

Balancer

Web ServerLoad

BalancerWeb ServersWeb Servers FirewallFirewall

App ServerLoad

Balancer

App ServerLoad

BalancerApplication

ServersApplication

ServersDatabaseCluster

DatabaseClusterFirewallFirewall

FirewallsLoad

Balancers Network Servers Storage



Grid Grid Grid


Grid Component Life Cycle -Example 1 - Server

ProvisionProvisionProvision DecommissionDecommissionDecommissionLogicallyManageable -Server is physically

installed withappropriate network

and storage links

PhysicallyManageable

Powered Up


Grid Component Life Cycle -Example 2 - Operating System

ProvisionProvisionProvision DecommissionDecommissionDecommissionOS Instance whichmay be bound andbooted on a server or

virtual machineinstance

Unconfigured OSinstance

OS instance bound toa server or virtualmachine manager,

booted and availablefor use


Grid Component Life Cycle -Example 3 - Database Service

ProvisionProvisionProvision DecommissionDecommissionDecommissionDatabase Service, incConfigured server and

Database tables

UnconfiguredDatabaseService

Database instancebound to an OS orOS partition andavailable for use


The Grid Management Entity (GME)• Manages the collection of Grid Components• Manages the relationships between them• Manages their life cycle• May be realized

– Through people and process– Through automated tools– A combination of the above

• Provides the context for understanding the key problems associated with managing Grid Components today

• Is the Macro-Actor in all use cases


EGA Reference Model:Role of Grid Management Entity

Manage Monitor

Accounting &BillingPolicyNotifications,Patches etc.

Managing includesProvisioning

DiscoveryConfigurationActivation

ControllingStatePerformanceScalingAvailabilitySecurity

Decommissioning

Monitoring includesStatePerformanceScalingAvailabilitySecurityAccounting

GMEGME

GridComponent

GridComponent

EnterpriseEnterpriseVendorSupportVendorSupport

Bugs etc.

Other GMEsOther GMEs


EGA Reference Model:GME and Grid Component Life Cycle

Discover,Control,Monitor,Notify

GMEGME


EGA Reference Model:GME and Service Level Management

Manage Monitor



Assigns/Provisions

Consumed Generates

ReconciledGMEGME

MetricsMetricsResources

(other Grid Components)

Resources(other

Grid Components)

GridComponent

GridComponent

Bugs etc.


Grid Management Entity

App ServerInstance

App ServerInstance

App ServerInstance

App ServerInstance

PersistentStorage

Tier

App ServerCluster

Policy Delegation

Embedded GMEFunctionality InCluster ManagementFramework

GMEGMEGME

DistributedHierarchical


EGA Reference Model: GME Delegated Policy and SLM

VendorSupportVendorSupport EnterpriseEnterprise

ElectronicBookstore

GridComponent

ElectronicBookstore

GridComponent

Data CenterGME

Component

Data CenterGME

Component

BookstoreGME

Component

BookstoreGME

Component

DatabaseCluster

GridComponent

DatabaseCluster

GridComponent

Application ServerCluster

GridComponent

LogicalGME

DerivedPolicy

Accounting& Billing

BusinessPolicy

Bugs etc.NotificationsPatches etc.

ProvisionAccounting& Billing

Manage

Monitor

DatabaseCluster

GMEComponent

DatabaseCluster

GMEComponent

Accounting& Billing

Manage

Monitor

DerivedPolicy

Provision


GME Delegated Policy & SLMVendorSupport Enterprise

ElectronicBookstore

GridComponent

Data CenterGME

Component

BookstoreSLM GME

Component

DatabaseCluster

GridComponent

DatabaseCluster

GridComponent


GridComponent

LogicalGME

DerivedPolicy

Accounting& Billing

BusinessPolicy



Manage

Monitor

DatabaseCluster

SLM GMEComponent

DatabaseCluster

SLM GMEComponent

Accounting& Billing

Manage

Monitor

DerivedPolicy

Provision

Database Cluster Product


VendorSupportVendorSupport EnterpriseEnterprise

ElectronicBookstore

GridComponent

ElectronicBookstore

GridComponent

Data CenterGME

Component

Data CenterGME

Component

BookstoreGME

Component

BookstoreGME

Component

DatabaseCluster

GridComponent

DatabaseCluster

GridComponent


GridComponent

LogicalGME

DerivedPolicy

Accounting& Billing

BusinessPolicy



Manage

Monitor

DatabaseCluster

GMEComponent

DatabaseCluster

GMEComponent

Accounting& Billing

Manage

Monitor

DerivedPolicy

Provision

EGA Reference Model:GME Functional Breakdown

PolicyPolicy

Control��Control��

Provisioning/-Decommissioning�

Provisioning/-Decommissioning�

MonitorMonitor

AvailabilityManagement ��

AvailabilityManagement ��

PerformanceManagementPerformanceManagement

Security ��Security ��

Accounting & Billing

Accounting & Billing

OrchestrationOrchestration

ServiceLevelManagement


The Functional Breakdown Of The GME



Bugs etc.

Other GMEsOther GMEs


ServerServer

DatabaseInstance

DatabaseInstance

BootDisk/LUN

BootDisk/LUNOS MediaOS Media

DatabaseStorage

DatabaseStorage

OSPatches

OSPatches




DatabaseMedia

DatabaseMedia



OS InstanceGME

OS InstanceGME

ServerGME

ServerGME

DatabaseInstance

GME

DatabaseInstance

GME

BootDisk/LUN

GME

BootDisk/LUN

GMEOS Media

GMEOS Media

GME

DatabaseStorage

GME

DatabaseStorage

GME

OSPatches

GME

OSPatches

GME

OS ContainerGME

OS ContainerGME

DatabaseTx Log Storage

GME

DatabaseTx Log Storage

GME

DatabaseMediaGME

DatabaseMediaGME


GME


GME

EnterpriseGridGME

EnterpriseGridGME

LogicalGME


EGA Reference Model: Use Cases• Key to articulating requirements• Provide basis for collaboration• Overall approach

– Focused - prioritization driven by EGA working groups and their needs

– Scenarios drive the larger context– Use cases built in detail from the bottom up

Leverage ref model - DAGs, life cycle, GME functionality etc.Provide operational environment context

• Opportunity to ensure alignment with existing best practices such as ITIL & eTOM


EGA Roadmap: Technical Focus• Reference Model – v2.0

– More detailed use cases (~60-100), due later this year– Mapping onto other standard architectures and information models

• OGSA, CIM, ITIL

• Component Provisioning– Provisioning (server) use cases and requirements

• Data Provisioning– Usage scenarios and mapping of existing specs

• Grid Security– Enterprise grid security requirements– Issues and resolutions and use cases

• Utility Accounting– Currently exploring requirements


EGA Roadmap: It’s About Iteration• We need feedback• We encourage membership• We want to make it more useful


EGA Roadmap: Strategic Relationships• Developing strategic relationships

– Identify synergies– Build a long-term, cooperative foundation

• DMTF - active engagement/cooperation– Joint work register has been created

and approved by EGA and DMTF Boards– Publicly announced

• SNIA - developing relationship• GGF - defining relationship

That’s why I’m here!

enterprise grid alliance - open grid forum use cas… · enterprise grid alliance accelerating the...

Documents