Private CloudPrinciples, Concepts and Patterns

TomShinderPrincipal WriterSCD iX Solutions Group

Agenda• Why Architecture?• Cloud Service and Deployment Models• Private Cloud Principles, Concepts and Patterns• Top Ten Private Cloud Architecture Issues• Patterns in Infrastructure as a Service (IaaS)

Why Architecture?

Is Knowing Architecture Useful?

• “Architects are space cadets”• “Architects draw star charts”• “Architects don’t have demos or HOLs”• “Architects can’t do anything (don’t know how to do

anything)”• “I don’t know any architects”• “I’ve never needed an architect”• “What’s an architect?”

What’s in it for me?

CorporateExecutiveBoard

Gartner

Is this your network today?

Anatomy

Physiology

Pharmacology

Biochemistry

Neuroscience

Pathology

MicrobiologyDefinitionsConstraints

RequirementsDecision Points

Cloud Service and Deployment Models

NIST Cloud Definition

CommunityCloud

Private Cloud

Public Cloud

Hybrid CloudsDeploymentModels

ServiceModels

EssentialCharacteristics

Common Characteristics

Infrastructure as a Service (IaaS)

Platform as a Service (PaaS)

Software as a Service (SaaS)

Resource Pooling

Broad Network Access Rapid Elasticity

Measured Service

On Demand Self-Service

Low Cost Software

Virtualization Service Orientation

Advanced Security

Homogeneity

Massive Scale Resilient Computing

Geographic Distribution

Private CloudPrinciplesConceptsPatterns

The Ten Private Cloud Principles

Create A Seamless User

Experience

Optimize Resource Utilization

Ubiquitous Automation

Drive Predictability

Incentivize Desired Behavior

Holistic Approach to Availability

Perception of Infinite

Capacity

Service Providers Approach

Perception of Continuous Availability

Business ValueContinuous

Improvement

Principles provide general rules and guidelines to support the evolution of a cloud infrastructure. They are enduring, seldom amended, and inform and support the way a cloud fulfills its mission. They strive to be compelling and aspirational. These

principles form the basis on which a cloud infrastructure is planned, designed and created

The Fourteen Private Cloud Concepts

Service Classification

Consumption Based Pricing

Security and Identity

MultitenancyCost

Transparency

Virtualized Infrastructur

e

Elastic Infrastructur

e

Partition Shared

Resources

Resource Decay

Fabric Managemen

t

Resiliency over

Redundancy

Pool Compute Resources

Homogenized Hardware

PredictabilityConcepts are abstractions or strategies that support the principles and facilitate the composition of a cloud. They are guided by and directly support one or more of the principles.

The Nine Private Cloud Patterns

Service Class Cost ModelHealth Model

Scale Unit Capacity PlanReserve Capacity

Physical Fault Domain

Upgrade Domain

Resource PoolingPatterns are specific, reusable ideas that have been proven

solutions to commonly occurring problems. Patterns are useful for enabling the cloud computing concepts and principles.

Top Ten Private Cloud Architectural Issues

New World Order – New Ways of Thinking It’s a

best practice

!

Gartner says this is the way!

That’s how we’ve always done it!

Our auditors

said to do it this way!

How Do You Cool Your Datacenter?

Passive flow air exchanger

Guess what? You’re a Service Provider!

Principle #4

You’ve got competition

Requires service management

maturity

Approach to service

management needs to change

Virtualization is an Enabler – Not a Cloud Solution

Can Reduce Quality of Service

Greater Complexity More Reactive

Requires Right Monitoring

ToolsMTRS Goes Up

Homogeneity Drives Multiple Efficiencies

Simplicity is Elegance

• Drives predictable performance

• Reduces cost of acquisition• Help with predicting time for

new acquisition

Principle #3

Redundancy is so 1990s

Redundancy comes with a price

Capital Expense

Operational Expense (power, maintenance)

Stateful applications increase cost

Driven by stateful applications

Principle #2

Is Redundancy Worth the Cost?

• No UPS• No

Generator• Straight

Utility Power

• Availability: 0.999

• Failure Probability in 5 years: ~100%

Eco

nom

y C

lass • UPS

• No Generator:• Availability:

0.999 • Failure

Probability in 5 years: 87%

• With Generator:• Availability:

0.999• Failure

Probability in 5 years: ~50%B

usi

ness

Cla

ss

• N+1 UPS• N+1

Generator

• Availability: 0.999999

• Failure Probability in 5 years: <1%

Gulfst

ream

Fail Often – Recover Fast!

Software enabled resiliency plans for failure states and focuses on fast service restoration. More failures, but less total service

downtime

Private Cloud

Fabric Management Enables Resiliency

• Avoid hardware failure

• Redundant at all levels

• Longer MTBF – More Disruption

Redundancy Driven HA

• Goal – minimize service disruption

• Automated fault detection and response

• Fail often – recover fast!

Resiliency Driven HA

Principle #2

Transform Users into EnvironmentalistsPay only for

what you use

Show the cost of five nines

Elasticity – both up and down

Portable apps enable cost comparison

Concept #9

Take Your Hardware to the Doctor

Health Model must

provide visibility

into hardware

infrastructure

Pattern #7

More to services than software

Understand hardware

interactions and dependencies

What does “healthy” look

like?

What does “unhealthy” look

like?

What does “failure” look

like?

Detect and respond depend

on understanding

Is it a Headache or a Hemorrhage?Define both Failure and

Disaster

• How do failures affect the service as a whole?

• Create a dependency tree• Determine when failure becomes

disaster• Detect and automate both failure and

disaster responses

Pattern #7

Domo Arigato, Mr. Roboto

Automation Drives

the Cloud

Manual• Slow/Error Prone

Mechanized• Faster/Predictable

Automation• Fast and Predictable

Patterns in IaaS

Infrastructure as a Service

DDC Hypervisor Fabric

Health State

Management System

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Host Location Management

VirtualHost

VirtualHost

VirtualHost

VirtualHost

VirtualHost

VirtualHost

VirtualHost

VirtualHost

VirtualHost

VirtualHost

IaaS Overview

Resource Pool

Scale Unit

+Fault

Domain

Upgrade Domain

DDC Hypervisor Fabric

Health State

Management System

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Host Location Management

VirtualHost

VirtualHost

VirtualHost

VirtualHost

VirtualHost

VirtualHost

VirtualHost

VirtualHost

VirtualHost

VirtualHost

DDC Hypervisor Fabric

Health State

Management System

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Host Location Management

VirtualHost

VirtualHost

VirtualHost

VirtualHost

VirtualHost

VirtualHost

VirtualHost

VirtualHost

VirtualHost

VirtualHost

DDC Hypervisor Fabric

Health State

Management System

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Physical Server

Host Location Management

VirtualHost

VirtualHost

VirtualHost

VirtualHost

VirtualHost

VirtualHost

VirtualHost

VirtualHost

VirtualHost

VirtualHost

X

X

Resource Decay

Resource PoolPrinciple

#4

Scale Unit

+

Pattern #5

Standardized increments

Known environmental requirements

Known procurement lead time

Standardized engineering

Compute scale unit

Storage scale unit

Resource DecayPrinciple

#9

Move away from break/fix incident model

Use a pool-based maintenance model

Define % of decay before maintenance

Fault DomainPattern #2 Consider non server component failures

Select the component most likely to fail

How many servers are impacted?

That’s the physical fault domain

Upgrade DomainPattern #3

Host servers still need to be upgradedAll in UD are maintained simultaneouslyWorkloads migrated away during upgrade

Worst Case Scenario

3% in decay2% in upgrade

10% lost in fault domain15% of total capacity lost

Reserve CapacityPattern #4

Takes advantage of homogeneity

VMs can be relocated predictably

Combines decay/fault/upgrade concepts

Total is Reserve Capacity

RC=FD+D(max)+UP/RP

There should be no decrement in service

So, what do you think of architecture?

For More Information

www.technet.com/cloud/private-cloud

http://social.technet.microsoft.com/wiki/contents/articles/6642.a-solution-for-private-cloud-security.aspx

Free Stuff!

This presentation is yours!• Re-present it• Lots of speakers notes• Improve it!

QuestionsDiscussionNew Ideas

© 2012 Microsoft Corporation. All rights reserved. Microsoft, Windows, Windows Vista and other product names are or may be registered trademarks and/or trademarks in the U.S. and/or other countries.The information herein is for informational purposes only and represents the current view of Microsoft Corporation as of the date of this presentation. Because Microsoft must respond to changing market conditions, it should not be interpreted to be a commitment on the part of Microsoft, and Microsoft cannot guarantee the accuracy of any information provided after the date of this presentation. MICROSOFT MAKES NO WARRANTIES, EXPRESS, IMPLIED OR STATUTORY, AS TO THE INFORMATION IN THIS PRESENTATION.