book

Cloud Architecture Patterns

by Bill Wilder

September 2012

Intermediate to advanced

182 pages

5h 12m

English

O'Reilly Media, Inc.

Read now

Unlock full access

AudienceWhy This Book ExistsAssumptions This Book MakesContents of This BookBuilding Page of Photos on Windows AzureTerminologyConventions Used in This BookUsing Code ExamplesSafari® Books OnlineHow to Contact UsAcknowledgments
Scalability DefinedVertically Scaling UpHorizontally Scaling OutDescribing ScalabilityThe Scale UnitResource Contention Limits ScalabilityEasing Resource ContentionScalability is a Business ConcernThe Cloud-Native ApplicationCloud Platform DefinedCloud-Native Application DefinedSummary
ContextCloud SignificanceImpactMechanicsCloud Scaling is ReversibleCloud scaling terminologyManaging Session StateSession state varies by application tierSticky sessions in the web tierStateful node challengesSession state without stateful nodesStateless service nodes in the service tierManaging Many NodesEfficient management enables horizontal scalingCapacity planning for large scaleSizing virtual machinesFailure is partialOperational data collectionExample: Building PoP on Windows AzureWeb TierStateless Role Instances (or Nodes)Service TierOperational Logs and MetricsSummary
ContextCloud SignificanceImpactMechanicsQueues are ReliableProgramming Model for ReceiverInvisibility window and at-least-once processingIdempotent processing for repeat messagesPoison messages handling for excessive repeatsUser Experience ImplicationsScaling Tiers IndependentlyExample: Building PoP on Windows AzureUser Interface TierService TierSynopsis of Changes to Page of Photos SystemSummary
ContextCloud SignificanceImpactMechanicsAutomation Based on Rules and SignalsSeparate ConcernsBe Responsive to Horizontally Scaling OutDon’t Be Too Responsive to Horizontally Scaling InSet Limits, Overriding as NeededTake Note of Platform-Enforced Scaling LimitsExample: Building PoP on Windows AzureThrottlingAuto-Scaling Other Resource TypesSummary
CAP Theorem and Eventual ConsistencyEventual Consistency ExamplesRelational ACID and NoSQL BASEImpact of Eventual Consistency on Application LogicUser Experience ConcernsProgrammatic DifferencesSummary
ContextCloud SignificanceImpactMechanicsMapReduce Use CasesBeyond Custom Map and Reduce FunctionsMore Than Map and ReduceExample: Building PoP on Windows AzureSummary
ContextCloud SignificanceImpactMechanicsShard IdentificationShard DistributionWhen Not to ShardNot All Tables Are ShardedCloud Database InstancesExample: Building PoP on Windows AzureRebalancing FederationsFan-Out Queries Across FederationsNoSQL AlternativeSummary

MultitenancySecurityPerformance ManagementImpact of Multitenancy on Application LogicCommodity HardwareShift in Emphasis from MTBF to MTTRImpact of Commodity Hardware on Application LogicHomogeneous HardwareSummary
ContextCloud SignificanceImpactMechanicsTransient Failures Result in Busy SignalsRecognizing Busy SignalsResponding to Busy SignalsUser Experience ImpactLogging and Reducing Busy SignalsTestingExample: Building PoP on Windows AzureSummary
ContextCloud SignificanceImpactMechanicsFailure ScenariosTreat All Interruptions as Node FailuresMaintain Sufficient Capacity for Failure with N+1 RuleHandling Node ShutdownNode shutdown with minimal impact to user experienceNode shutdown without losing partially completed workNode shutdown without losing operational dataRecovering From Node FailureShielding interactive users from failuresResuming work-in-progress on backend systemsExample: Building PoP on Windows AzurePreparing PoP for FailureN+1 ruleWindows Azure fault domainsUpgrade domainsHandling PoP Role Instance ShutdownWeb role instance shutdownWorker role instance shutdownUse controlled rebootsRecovering PoP From FailureSummary
Network Latency ChallengesReducing Perceived Network LatencyReducing Network LatencySummary
ContextCloud SignificanceImpactMechanicsAutomation HelpsCost ConsiderationsNon-Technical ConsiderationsExample: Building PoP on Windows AzureAffinity GroupsOperational Logs and MetricsSummary
ContextCloud SignificanceImpactMechanicsPublic AccessGranting Temporary AccessSecurity ConsiderationsExample: Building PoP on Windows AzurePublic Read AccessShared Access SignaturesSummary
ContextCloud SignificanceImpactMechanicsCaches Can Be InconsistentExample: Building PoP on Windows AzureCost ConsiderationsSecurity ConsiderationsAdditional CapabilitiesSummary
ContextCloud SignificanceImpactMechanicsNon-Technical Considerations in Data Center SelectionCost ImplicationsFailover Across Data CentersExample: Building PoP on Windows AzureChoosing a Data CenterRouting to the Closest Data CenterReplicating User Data for PerformanceReplicating Identity Information for Account OwnersData Center FailoverColocation AlternativesSummary
Page of Photos (PoP) SampleResources From Preface and ChaptersChapter 1Chapter 2Chapter 3Chapter 4Chapter 5Chapter 6Chapter 7Chapter 8Chapter 9Chapter 10Chapter 11Chapter 12Chapter 13Chapter 14Chapter 15

Content preview from Cloud Architecture Patterns

Appendix A. Further Reading

Page of Photos (PoP) Sample

The Page of Photos sample application is used throughout the book. Parts of it have already been implemented using Windows Azure to demonstrate ideas in the book. The code for Page of Photos (minimalist at first, then built out over time by the author and some accomplices) will be shared in a public repo on GitHub.

Run the Page of Photos (PoP) sample application: http://www.pageofphotos.com
View the Page of Photos (PoP) sample application source code: http://www.github.com/codingoutloud/pageofphotos
Find information about the book and possibly related content in the future: http://www.cloudarchitecturebook.com

Resources From Preface and Chapters

Windows Azure Platform: http://www.windowsazure.com
Amazon Web Services: http://aws.amazon.com/
Google App Engine: http://developers.google.com/appengine/
A NIST Definition of Cloud Computing (SP 800-145 Sept. 2011): http://csrc.nist.gov/publications/nistpubs/800-145/SP800-145.pdf
(NIST) Cloud Computing Synopsis and Recommendations (SP 800-146 May 2012): http://csrc.nist.gov/publications/nistpubs/800-146/sp800-146.pdf

Chapter 1

“A Compuware analysis of 33 major retailers across 10 million home page views showed that a 1-second delay in page load time reduced conversions by 7%.” Source: Compuware, April 2011.
“Google observed that adding a 500-millisecond delay to page response time caused a 20% decrease in traffic” Source: Marissa Mayer, “What Google Knows” talk at Web 2.0 Conf 2006 (11/09/2006): ...