book

High Availability: Design, Techniques, and Processes

by Floyd Piedad, Michael Hawkins

December 2000

Intermediate to advanced

288 pages

5h 45m

English

Pearson

Read now

Unlock full access

Complexity, Complexity, ComplexityMultiple Technologies and ProtocolsMultiple VendorsVaried UsersMultiple LocationsRapid ChangeGreater Business DemandsA Daunting Environment To Work InThe Total Cost of Ownership IssueTotal Cost of Ownership DefinedIndustry TCO EstimatesWhat TCO Studies RevealThe Underlying Reason for High TCOA Typical Scenario: Choosing Office SystemsAvailability as the Most Significant Contributor to TCOSummary
Determining User Availability RequirementsThe Service Level AgreementHelping Users Identify Their Availability RequirementsAvailability Levels and MeasurementsHigh Availability LevelContinuous Operations LevelContinuous AvailabilityQuantifying Availability TargetsAvailability: A User MetricMeasuring End-To-End AvailabilitySummary
Identifying System ComponentsAddressing Critical ComponentsThe Four Elements of AvailabilitySummary
Processes, Data, Tools, and OrganizationSystems Management in the PC World (or the Lack of It)IT Organizations: Away from Centralization, Then Back AgainUnderstanding the Systems To ManageThe Basics of Management: Five PhasesPhase 1: Setting ObjectivesPhase 2: PlanningPhase 3: ExecutionPhase 4: MeasurementPhase 5: ControlIdentifying the Systems Management Disciplines
Service-Level ManagementProcess RequirementsStep 1: Define service-level standardsStep 2: Establish service levels to be attainedStep 3: Monitor achievement of service levelsStep 4: Analyze service-level attainment and report to higher managementStep 5: Redefine service levels, if necessaryData and Measurement RequirementsOrganization RequirementsTools RequirementsBenefits of Service-Level ManagementProblem ManagementProcess RequirementsStep 1: Define problem management process and practicesStep 2: Detect or recognize the problemStep 3: Bypass the problemStep 4: Analyze the problemStep 5: Manage the problem to resolutionStep 6: Report on the status and trends of problemsStep 7: Redefine the problem management process if necessaryData RequirementsOrganization RequirementsTools RequirementsBenefits of Problem ManagementChange ManagementProcess RequirementsStep 1: Define change management process and practicesStep 2: Receive change requestsStep 3: Plan for implementation of changesStep 4: Implement and monitor the changes, back out changes if necessaryStep 5: Evaluate and report on changes implementedStep 6: Modify change management plan if necessaryData RequirementsOrganization RequirementsTools RequirementsBenefitsSecurity ManagementProcess RequirementsStep 1: Determine and evaluate of IT assetsStep 2: Analyze riskStep 3: Define security practicesStep 4: Implement security practicesStep 5: Monitor for violations and take corresponding actionsStep 6: Reevaluate IT assets and risksData RequirementsOrganization RequirementsTools RequirementsBenefitsAsset and Configuration ManagementProcess RequirementsStep 1: Define asset and configuration data requirementsStep 2: Identify asset and configuration information gathering and update proceduresStep 3: Gather asset and configuration information and update proceduresStep 4: Provide information to other systems management disciplinesStep 5: Analyze asset and configuration information qualityStep 6: Reevaluate asset and configuration data and their update requirementsData RequirementsOrganization RequirementsTools RequirementsAvailability ManagementProcess RequirementsStep 1: Define a plan to achieve availability targetsStep 2: Track availability targets and their achievementStep 3: Analyze and report on availability achievementsStep 4: Update availability planData RequirementsOrganization RequirementsTools RequirementsBenefits
Systems Management DisciplinesThe Centralized Computing EnvironmentThe Distributed Computing EnvironmentSystems Management in Today's Computing EnvironmentDefining Appropriate Functions and ControlCentralized management and controlCentralized management, distributed controlDistributed management and controlChoosing a Deployment StrategyCapability to manage remote resourcesSkills availability at remote locationsPerformance impact of managing remote resources centrallyThe need for greater security and controlPhysical proximity of resources to each otherDeveloping a Deployment StrategyManagement by ExceptionPolicy-Based ManagementStandardization of Performance DataAccountability of the Distributed Systems ManagerCentral Definition of Systems Management ArchitecturesProcess OwnershipSummary

RedundancyHardware Redundancy ExamplesSoftware Redundancy ExamplesEnvironmental Redundancy ExampleCritical Success FactorsBackup of Critical ResourcesMethods of BackupHardware Backup ExamplesSoftware Backup ExamplesIT Operations Backup ExamplesCritical Success FactorsCurrency of backupAutomated updating of backupsIsolation of backup from primaryBackup and restore procedure review and testingGenerations of backupsIntegrity verificationClusteringComparing Clustering and RedundancyHardware and Software Clustering ExamplesIT Operations Clustering ExamplesEnvironmental Clustering ExamplesCritical Success FactorsAutomatic load-sharingPhysical separation of clustered componentsFault ToleranceHardware Fault Tolerance ExamplesSoftware Fault Tolerance ExamplesEnvironmental Fault Tolerance ExamplesCritical Success FactorsIsolation or PartitioningHardware Isolation ExamplesSoftware Isolation ExamplesOther Benefits of IsolationMinimize risk of changesReduce resource contentionMaximize resourcesSimpler systems management proceduresCritical Success FactorsAutomated OperationsConsole and Network Operations ExamplesWorkload Management ExamplesSystem Resource Monitoring ExamplesProblem Management ApplicationsDistribution of Resources ExampleBackup and Restore ExamplesCritical Success FactorsAccess Security MechanismsSteps to Secure AccessStep 1: Identify the person requesting accessStep 2: Verify the identityStep 3: Control accessStep 4: Monitor all activitiesTypes of SecurityPhysical securityNetwork securityApplication securityComputer resource securityPassword ManagementStep 1: Enforce password selection guidelinesStep 2: Expire passwords regularlyStep 3: Expire assigned passwords on first useStep 4: Disable user accounts after successive invalid password attemptsStep 5: Educate users on how to protect their password informationCritical Success FactorsStandardizationHardware Standardization ExamplesSoftware Standardization ExamplesNetwork Standardization ExamplesProcesses and Procedures Standardization ExamplesNaming Standardization ExamplesCritical Success FactorsTransitioning to StandardizationSummary
The Use of Reliable ComponentsTechniques for Maximizing Hardware Component ReliabilityChoose components with low failure ratesChoose components that have high MTBFPurchase from reputable suppliersUse technical specifications as a gaugeChoose products with fewer parts or greater integrationAvoid newly developed products whenever possibleFollow maintenance schedules diligentlyTechniques for Maximizing Software Component ReliabilityAvoid using "Version 1" and "Beta" softwareDon't use shareware or freewareBuy industry-standard software from reliable vendorsPrior to installation, test for virusesProvide menus and other ways to control user inputsReuse bug-free components or modulesTest programs thoroughlyRun "beta tests" with a controlled set of usersInstall the latest application software fixes judiciouslyInstall the latest device drivers when availableUpgrade to newer operating systems with cautionMinimize the use of system utilitiesPersonnel-Related Techniques for Maximizing ReliabilityEnsure high-quality user trainingEnsure quality training of support staff membersBe wary of contractual hiresEnvironment-Related Techniques for Maximizing ReliabilityInstall Automatic Voltage Regulators (AVRs)Use adequate air-conditioning equipmentSome Reliability Indicators for SuppliersTime in businessQuality certificationIndustry awardsPeer recommendationWarranty and supportProgramming to Minimize FailuresCorrectnessEnsure user requirements are adequately determinedPrototype the application prior to detailed codingRevalidate user requirements midway through the projectBeta test prior to wide-scale deploymentRobustnessTest against out-of-bounds valuesTrap errors and prevent them from propagatingAnticipate external changesExtensibilityUser changesSystem platform changesRegulatory changesBudgetary changesBusiness volume changesBusiness demand changesGenerous database field sizesDesign with overcapacityPlace constant values in a look-up tableReusabilityImplement Environmental Independence MeasuresUse Power GeneratorsUse Independent Air-Conditioning UnitsUse Fire Protection SystemsUse Raised FlooringInstall Equipment Wheel LocksLocate Computer Room on the Second FloorUtilize Fault Avoidance MeasuresAnalyzing Problem Trends and StatisticsUse of Advanced Hardware TechnologiesUse of Software Maintenance ToolsSummary
Automatic Fault RecognitionParity Checking MemoryECC MemoryData Validation RoutinesFast Recovery TechniquesMinimizing Use of Volatile Storage MediaRegular Database Updates to Central StorageAutomatic File-Save FeaturesSummary
Online System RedefinitionAdd or Remove I/O DevicesSelectively Power Down SubsystemsCommit or Reject ChangesInformative Error MessagesUse Standard Corporate TerminologyAdopt Terms Already Used by Common ApplicationsTell What, Why, Impact, and HowImplement Context-Sensitive HelpGive Options for Viewing More Detailed Error InformationMake Error Information Available After the Error Has Been ClearedComplete DocumentationHave a Manual of Operations On HandWrite Basic Problem Isolation and Recovery GuidesProvide System Configuration DiagramsLabel ResourcesProvide a Complete Technical LibraryInstallation of Latest Fixes and PatchesSummary
Use Manageable ComponentsSimple Network Management Protocol (SNMP)Common Management Information Protocol (CMIP)Desktop Management Interface (DMI)Common Information Management Format (CIM)Wired for Management (WfM)Management ApplicationsSystems Management IssuesDeploymentOperationsSecurityAutomated Systems Management CapabilitiesSystem Management Applications and FrameworksUnicenter TNG (Computer Associates)Tivoli (IBM)Educate IS Personnel on Systems Management DisciplinesBusiness Value of the Information SystemValue of Systems Management DisciplinesPrinciples of ManagementBasic Numerical Analysis SkillsSummary
The Value of Systems Management DisciplinesWhich One First?Analyze OutagesIdentify Single Points of FailureExploit What You HaveAn Implementation StrategySummary
Availability Features of Selected Operating SystemsAvailability Features of Novell NetWareSFT II (System Fault Tolerance level 2) or mirroring/duplexingSFT III (System Fault Tolerance level 3)Dynamic load/unloadClient auto-reconnectKernel fault recovery or ABEND recoveryNovell Replication Services (NRS)Novell Application Launcher (NAL)Hot plug PCIMultiprocessor Kernel (MPK)Intelligent I/O (I2O)Memory protectionFlexible Mirroring, Phase INovell Storage Services (NSS)Hierarchical Storage Management (HSM)NetWare Cluster Services (NWCS)Availability Features of Sun Solaris 8Sun ClusterSolaris Resource Manager/Solaris Bandwidth ManagerDynamic Reconfiguration/Automated Dynamic ReconfigurationNetwork multipathingLive upgradesHot patching for diagnosticsImproved crash dump analysisImproved program analysisBetter examination of core filesBus performance monitoringBetter management of core filesImproved device configurationMacro-level debuggingRemote console messagingTCP/IP network diagnosticsIP packet routing observabilitySystem crash dump utilityEnhanced process tracingAvailability Features of AIXLogical Volume Manager (LVM)Disk mirroringBad block relocationJournaled File System (JFS)Dynamic AIX kernelHigh Availability Cluster Multiprocessing (HACMP)System Resource Controller (SRC)Configuration managerAIX update facilitiesAvailability Features of Microsoft Windows 2000 Server and ProfessionalWindows File ProtectionDriver certificationKernel-mode write protectionIIS application protectionCluster services and Network Load Balancing (Advanced Server and Datacenter Server)Job object API and process controlApplication certification & DLL protectionDistributed File System (Dfs)Disk quotasHierarchical Storage Management (HSM)Rolling upgrade support (Advanced Server and Datacenter Server)Dynamic Volume ManagementError handling and protected subsystemsAutomatic restartKill process treeSystem preparation toolWindows InstallerPlug and Play (PnP)Service pack slipstreamingIntegrated directory services (Active Directory)Windows Management Instrumentation (WMI)Delegated administrationMicrosoft Management Console (MMC)Windows Script Host (WSH)Group policies and centralized desktop managementRecoverable file systemDisk mirroring (RAID Level 1)Disk duplexingDisk striping with parity (RAID Level 5)Availability Features of IBM OS/400Policy-based Backup Recovery and Management Services (BRMS)Commitment control and journalingSystem Managed Access Path Protection (SMAPP)Availability Features of Selected Hardware ComponentsAvailability Features of IBM S/390 Integrated ServerHigh availability storage devicesSSA Disk: Serial Storage Architecture (SSA)Power managementAvailability Features of the IBM AS/400 Midrange SystemLogical partitions (LPAR)Disk failure recoveryPower utility failure recoveryContinuously Powered Main (CPM) storageOperations NavigatorAS/400 system clusteringAvailability Features of the IBM RS/6000Built-in error detection and correctionBackup power supplyBattery backup systemsRedundant or spare disksHot-pluggable disk drivesMulti-tailed disks and shared volume groupsRAID disk arraysAvailability Features of Compaq Proliant ServersECC memorySMART-2 Array Controller technologyRedundant network interface cardsUninterruptible Power SupplyCompaq Insight ManagerStandby Recovery ServerOn-Line Recovery ServerRapid recoveryRedundant power supplyOff-line backup processorExternal storageAvailability Features of Selected Software ComponentsAvailability Features of the Oracle8i DatabaseCache Fusion clusteringFast-Start architectureOnline reorganizationSingle system view

Content preview from High Availability: Design, Techniques, and Processes

About the Authors

Floyd Piedad is President of AKAsia Services Corporation, a leading Internet services company based in Manila, the Philippines.

Michael Hawkins is an IT management consultant who specializes in implementing and managing large IT infrastructures for companies throughout Asia and North America.