VMware vSphere 6.7 Clustering Deep Dive

ClusterDeepDive-Boek-9-manuscript-rubrik-vmug
ebook-Rubrik-org
	VMware
	VMware vSphere 6.7 Clustering Deep Dive
	Copyright © 2018 by Frank Denneman, Duncan Epping and Niels Hagoort
	ABOUT THE AUTHORS
	INTRODUCTION AND ACKNOWLEDGEMENTS
	Figure 1: Photo from 2012
	Frank Denneman, Duncan Epping, Niels Hagoort
	FOREWORD
	Chris Wahl
	Chief Technologist
	P1
	HIGH AVAILABILITY
	INTRO TO VSPHERE HIGH AVAILABILITY
	Figure 2: vSphere HA Concept
	Figure 3: VM and Application Monitoring
		vSphere 6.7
			What’s New?
			What is Required for HA to Work?
			Prerequisites
			Firewall Requirements
			Configuring vSphere High Availability
	Figure 4: vSphere HA Configuration
	COMPONENTS OF HA
	Figure 5: vSphere HA Components
	Figure 6: FDM.log
		HOSTD Agent
		vCenter
	Figure 7: VM Protection Status
	FUNDAMENTAL CONCEPTS
		Master Agent
			Election
	Figure 8: vSphere HA State - Master
	Figure 9: vSphere HA Files
		Slaves
	Figure 10: vSphere HA State - Slave
		Files for Both Slave and Master
			Remote Files
			Local Files
	Figure 11: vSphere HA Local Files
	Figure 12: PrettyPrint.sh Hostlist Example
		Heartbeating
		Network Heartbeating
		Datastore Heartbeating
	Figure 13: Datastore Heartbeating
	Figure 14: Datastore Heartbeating Selected
	Figure 15: Heartbeat File
		Isolated versus Partitioned
	Figure 16: Isolation vs Partition
		VM Protection
	Figure 17: VM is Protected
	Figure 18: VM Protection Workflow
	Figure 19: VM Unprotection
	RESTARTING VIRTUAL MACHINES
		Restart Priority and Order
	Figure 20: Restart Priority
	Figure 21: Start Next Priority VMs Batch When …
	Figure 22: Section of PrettPrint.sh Output
	Figure 23: Application Group
	Figure 25: VM to VM Rule Definition
	Figure 26: VM Dependency Restart Condition
		Restart Retries
	Figure 27: Restart Retry Timeline
		Failed Host
			The Failure of a Slave
	Figure 28: Restart Timeline for Slave Failure
		The Failure of a Master
	Figure 29: Restart Timeline for Master Failure
		Isolation Response and Detection
			Isolation Response
	Figure 30: Isolation Response Configuration
	Figure 31: VMs Overrides – Host Isolation Response
		Isolation Detection
			Isolation of a Slave
	Figure 32: Poweroff File
	Figure 33: Isolation Declared
	Figure 34: Restart of VM After Isolation
		Isolation of a Master
			Additional Checks
	Figure 35: Isolation Address
		Selecting an Additional Isolation Address
		Isolation Policy Delay
		Restarting VMs
		VM Component Protection
	Figure 38: Web Client PDL Response
	Figure 39: vSphere Client PDL Response
	Figure 40: vSphere Client APD Response
		vSphere HA respecting Affinity Rules
	Figure 41: vSphere HA Rule Settings
	VSAN AND VVOL SPECIFICS
		HA and vSAN
	Figure 42: vSAN Datastore
	Figure 43: vSAN Network RAID
		Folder Structure with vSAN for HA
	Figure 44: vSAN Partition Scenario
	Figure 45: vSAN Stretched Cluster
		Heartbeat Datastores, When Can They Help?
		HA and Virtual Volumes
	Figure 46: Virtual Volumes Enabling Capabilities
	Figure 48: Virtual Volumes Architecture
	Let’s take a look at all of these three in the above order.
	Figure 49: Virtual Volumes Folder Structure
	Figure 50: Virtual Volumes Object for HA Files
	ADDING RESILIENCY TO HA
	Cons: Just a single active path for heartbeats.
	Figure 52: Network Configuration
		Corner Case Scenario: Split-Brain
			Link State Tracking
	Figure 53: Link State Tracking
	ADMISSION CONTROL
		Admission Control Policy
	Figure 54: Admission Control
	Figure 55: Admission Control Algorithms
		Admission Control Algorithms
		Cluster Resource Percentage Algorithm
	Figure 56: Percentage Based
	Figure 57: Host Failures Cluster Tolerate
	So how does the admission control policy work?
	Figure 58: Admission Control Monitoring
		Slot Size Algorithm
	Figure 59: Host Failures Cluster Tolerates
	Figure 60: Slot Policy
	Let’s dig in to this concept we have just introduced, slots.
	Figure 61: Advanced Runtime Info
	Figure 62: VM Spanning Multiple Slots
	Figure 63: Change Memory Slot Size
	Figure 64: VMs Requiring Multiple Slots
	Figure 65: Large Memory Reservation
	Figure 66: Slots Available Changed
		Unbalanced Configurations and Impact on Slot Calculation
	Let’s first define the term “unbalanced cluster.”
	Let’s try to clarify that with an example.
	Figure 67: Unbalanced
	ESXi-01 + ESXi-02 = 32 slots available
		Failover Hosts
	Figure 68: Dedicated Failover Hosts
	Figure 69: Configure Failover Host Admission Control Policy
		Performance Degradation
	Figure 70: Performance Degradation Tolerated
		Decision Making Time
			Percentage as Cluster Resources Reserved
	Pros:
	Cons:
		Slot Size Algorithm
	Pros:
	Cons:
		Specify Failover Hosts
	Pros:
	Cons:
		Recommendations
			Selecting the Right Percentage
	A total of 112 GB of memory is reserved as failover capacity.
	Figure 71: Determining the Percentage
		Aggressive Approach
		Adding Hosts to Your Cluster
	Figure 72: Adding Hosts
	VM AND APPLICATION MONITORING
	Figure 73: VM and Application Monitoring
		Why Do You Need VM/Application Monitoring?
		How Does VM/App Monitoring Work?
	Figure 74: Monitoring Sensitivity
	Align das.iostatsInterval with the failure interval.
		Screenshots
		VM Monitoring Implementation Details
			Timing
		Application Monitoring
	Figure 75: Application Monitoring
		Application Awareness API
	VSPHERE HA INTEROPERABILITY
		HA and Storage DRS
		Storage vMotion and HA
		HA and DRS
			HA and Resource Fragmentation
			Flattened Shares
	Figure 76: Flatten Shares Starting Point
	Figure 77: Flattening of Shares
	Figure 78: Flattening of Shares
		DPM and HA
		Proactive HA
	Figure 79: Proactive HA
	Figure 80: Proactive HA Automated Remediation
	Figure 81: Proactive HA Provider
	ADVANCED SETTINGS
		How Do You Configure these Advanced Settings?
			Cluster Level
			FDM Host Level
			vCenter Level
		Most Commonly Used
	P2
	VSPHERE DISTRIBUTED RESOURCE SCHEDULER
	INTRODUCTION TO VSPHERE DRS
		Requirements
		Cluster Level Resource Management
		DRS Cluster Settings
	Figure 82: New DRS Cluster
		DRS Automation Levels
	Figure 83: DRS Automation Level
		Manual Automation Level
		Partially Automated Level
	Figure 84: DRS Metrics on Cluster Summary Screen
	Figure 85: Pending DRS Recommendations
		Fully Automated Level
		Per-VM Automation Level
	Figure 86: VM Override Options
	Figure 87: DRS Recommendation of Partially Automated VM
		Impact of Automation Levels on Procedures
		Initial Placement
			Old Initial Placement Behavior
			vSphere 6.7 Initial Placement Behavior
			Virtual Machine Performance Modelling
		vCenter Sizing
	Figure 88: VCSA Last Quarter Memory Utilization View
		DRS Thread per Cluster
	Figure 89: DRS Components
		Separate VDI Workloads From VSI Workloads
		Cluster Sizing
	The number of virtual machines versus the number of LUNs required:
		Supporting Technology
		vMotion
			Multi-NIC vMotion
			10/25/40GbE vMotion Network
			CPU Consumption of vMotion Process
			Encrypted vMotion
	Figure 90: Encrypted vMotion CPU Overhead on the Source Host
	Figure 91: Encrypted vMotion CPU Overhead on the Destination Host
	Figure 92: VM Options Encrypted vMotion
		CPU Headroom
	Enhanced vMotion Capability
		How Does EVC Work?
	Figure 93: EVC Compatibility Check
		Will EVC Impact Application Performance?
		Enabling and Disabling EVC
		Power Off VM Instead of Reboot
		EVC Requirements
		Conclusion
	RESOURCE DISTRIBUTION
		DRS Dynamic Entitlement
		Resource Scheduler Architecture
	Figure 94: DRS and Host-Local Schedulers
		DRS Scheduler
	Figure 95: Resource Pool Structure Mapped to ESXi Host-Local RP Tree
	Figure 96: Mapping Resource Pool Structure Across Host-Local RP Trees
		Local Scheduler
		Dynamic Entitlement Target
	Figure 97: Dynamic Entitlement Target
	Figure 98: Active - Consumed - Configured Memory
	Figure 99: Idle Consumed Memory Calculation
		Memory Metric for Load Balancing Enabled
	Figure 100: Memory Metric for Load Balancing Enabled
		Resource Contention
			DRS Dynamic Entitlement versus Host-Local Entitlement
		Resource Allocation Settings
	Figure 101: Resource Allocation Settings and Dynamic Entitlement
		Reservation
	Figure 102: 6 GB Minimum Entitlement
		Resource Pool-Level Reservation Behavior
		Virtual Machine-Level Reservation Behavior
		Admission Control and Dynamic Entitlement
		Shares
			Relative Priorities
	Figure 103: Parent, Child, Sibling Relationship Mapping
		CPU Shares
	MHzPerShare = MHzUsed / Shares
	Figure 104: Order of Priority
	Figure 105: VM02 Claiming Resources Up to its Reservation
		Memory Shares
		Calculating MinFree
		Memory State Transition Points
	Figure 106: Memory Transition Points Host with 256 GB Memory
		Memory Reclamation Techniques per State
		Share-Per-Page
		Resource Contention
	Figure 107: Dynamic Entitlement to Determine Reclamation
	Figure 108: Reclamation and Level of Contention
		Worst Case Allocation
		Limits
			Why Use Limits?
		Tying it All Together
	RESOURCE POOLS AND CONTROLS
		Root Resource Pool
	Figure 109: HA Disabled on DRS Cluster
	Figure 110: Cluster HA Failover Resources
		Resource Pools
	Figure 111: Resource Providers and Consumers
		Inflating or Deflating Resource Pools
		Host-Local Resource Pools
		Dividing Resources
		Resource Pools Are Not Folders
		Resource Pool Tree Structure
		Worst-Case Scenario Should Not Mimic Your Cluster Operational State
	Figure 112: DRS Entitlement Viewer
		Resource Pool Resource Allocation Settings
		Shares
	Figure 113: Resource Pool 1 and 2 Share Ratio
		Resource Pool Size
		Sibling Rivalry
	Figure 114: Sibling Rivalry between VM and Resource Pool
	Figure 115: First Level of Resource Distribution
	Figure 116: Second Level of Resource Distribution
	Figure 117: Additional Workload Introduced in Resource Pool RP-1
	Figure 118: Dynamic Entitlement
	Figure 119: Sibling Rivalry Within RP with Idle and Active VMs
		Share Levels are Pre-sets, not Classes
		The Resource Pool Priority-Pie Paradox
	Figure 120: VM Dynamic Entitlement Based on RP Share Value
		From Resource Pool Setting to Host-Local Resource Allocation
	Figure 121: Single Resource Pool Configuration
	Figure 122: Host-Local Resource Pool Mapping
	Figure 123: Multiple Resource Pool Configuration
	Figure 124: Resource Pool Mapping ESXi-01
	Figure 125: Cluster Resource Allocation Overview
		Resource Pool-Level Reservation
	Figure 126: Distribution of Resource Pool Reservation at 08:00 (8 AM)
	Figure 127: Distribution of Resource Pool Reservation at 11:00 (11 AM)
	Figure 128: Distribution of Resource Pool Reservation at 19:00 (7 PM)
		Child-Object-Level Reservation Inside a Resource Pool
	Figure 129: VM Reservation Within a Resource Pool
		Activation of Reservation
		Memory Overhead Reservation
	Static Overhead
	Dynamic Overhead
		Memory Overhead Reservation Appears as Resource Pool Reservation
	Figure 130: Memory Overhead Reservation
		Right Size Virtual Machines
		Expandable Reservation
	Figure 131: Reservation Type Fixed
	Figure 132: Power On Failure
	Figure 133: Power On Decision Workflow
		Traversing the Parent Tree
	Figure 134: Traversing the Parent Resource Pool Tree
		Reservations are Not Limits
	Figure 135: Reservation and Shares
		Resource Pool Limit
			Limits, Reservations and Memory Overhead Reservation
			Expandable Reservation and Limits
	CALCULATING DRS RECOMMENDATIONS
		When is DRS Invoked?
		Recommendation Calculation
			Constraints Correction
		Imbalance Calculation
			Current Host Load Standard Deviation
			Target Host Load Standard Deviation
			DRS Migration Threshold
	Figure 136: Migration Threshold
	Figure 137: DRS Migration Recommendation Workflow
		GetBestMove
	GetBestMove() {
		Cost-Benefit and Risk Analysis Criteria
		Cost
		Benefit
		Risk
	Combining the Cost-Benefit Risk
		MinGoodness
		Calculating the Migration Recommendation Priority Level
	Level 1 (conservative)
	Level 2 (moderately conservative)
	Level 3 (moderate)
	Level 4 (moderately aggressive)
	Level 5 (aggressive)
		Pair-Wise Balancing Thresholds
	Figure 138: Pair-Wise Balancing Threshold
		Network Aware DRS
			Initial Placement Enhancement
			Load-Balancing Enhancement
			Network Saturation Threshold
	Figure 139: Network-Aware DRS Thresholds on Logically Separated NICs
	Figure 140: Network-Aware DRS Thresholds 10 GbE NICs Configuration
		Advanced Setting
	IMPACTING DRS RECOMMENDATIONS
		DRS Additional Options
	Figure 141: DRS Cluster Additional Options
		VM Distribution
	Figure 142: VM Distribution
		Memory Balancing in Non-Overcommitted Clusters
	Figure 143: DRS Memory Metric for Load Balancing
		Out-of-the-box DRS Behavior
	Figure 144: Active, Consumed and Configured Memory
	Figure 145: Default Dynamic Entitlement Calculation
		Memory Metric for Load Balancing Enabled
	Figure 146: 100% Idle Consumed Memory
	Figure 147: Cluster Memory Utilization Consumed View
	Figure 148: Cluster Balanced State
	Figure 149: Sum of VM Memory Utilization Based on Active Memory
	Figure 150: Sum of VM Memory Utilization Based on Consumed Memory
	Figure 151: Memory Metric for Load Balancing Enabled
		CPU Over-Commitment (DRS Additional Option)
	Figure 152: Setting the CPU Over-Commit Ratio 4:1
		Maximum vCPUs per CPU Core
	Figure 153: MaxVcpusPerCore Advanced Option
		Maximum vCPU Per Cluster Percentage
	Figure 154: Web Client Option
	Figure 155: MaxVcpusPerClusterPct
		AggressiveCPUActive
		VM Size and Initial Placement
		MaxMovesPerHost
		Placement Rules
			VM-VM Affinity Rules
	VM-VM Anti-Affinity Rules
	VM-VM Affinity Rules – impact on HA
	VM-VM Affinity Rules – impact on DRS
		VM-Host Affinity Rules
	Figure 156: VM to Host Affinity Rule
		Should Run - Preferential Rules
		Must Run - Mandatory Rules
	Figure 157: vMotion Compatibility Check
	Figure 158: Compatibility Subset
		Rule Behavior
		Backup Your Affinity Rules
		VM Overrides
	Figure 159: VM Overrides
	Figure 160: DRS Automation Level
		Disabled Automation Level
		Manual Automation Level
		Partially Automation Level
		The Impact of DRS Automation Levels on Cluster Load Balance
			Disabled versus Partially and Manual Automatic Levels
			Risk versus Reward
	DISTRIBUTED POWER MANAGEMENT
	Figure 161: Enable DPM
	Figure 162: DPM Automation Level
	Figure 163: DPM Host Power Management Option
		Calculating DPM Recommendations
			Evaluating Resource Utilization
	Figure 164: Power Operations Regarding to Host Utilization Levels
	Advanced options
	Figure 165: DRS Cluster Advanced Options
		Historical Period of Interest
		Evaluating Power-On and Power-Off Recommendations
		Power-Off Recommendations
	Rejection of Host Power-Off Recommendations
		DPM Power-Off Cost/Benefit Analysis
		Power-Off Cost and Benefit Analysis Calculation
		Host Selection for Power-On Recommendations
		Host Power-On Recommendations
	Use homogeneous clusters, as DPM operates more efficiently.
		Impact of Advanced Settings on Host Power-On Recommendations
		Recommendation Classifications
			Priority Levels
			Power-Off Recommendations
			Power-On Recommendations
		Guiding DPM Recommendations
			DPM Standby Mode
			DPM WOL Magic Packet
			Baseboard Management Controller
			Protocol Selection Order
	P3
	VSPHERE STORAGE DRS
	INTRODUCTION TO VSPHERE STORAGE DRS
		Resource Aggregation
		Initial Placement
		Load Balancing
	Figure 166: Storage DRS Automation Level
	Figure 167: I/O Metric for Storage DRS Recommendations
		Affinity Rules
		Datastore Maintenance Mode
		Requirements
	STORAGE DRS INITIAL PLACEMENT
		User Interaction
	Figure 168: Selecting Storage During VM Creation Process
		Affinity Rules
	Figure 169: Default VM Affinity Rule
		Cluster Automation Level
	Figure 170: Cluster Automation Level
		DRS Mobility and Datastore Connection
		Space and I/O Load Consideration
		Space Utilization Threshold
	Figure 171: Datastore Space Utilization Threshold
		Datastore Cluster Defragmentation
		Depth of Recursion
	Figure 172: Depth of Recursion
		Goodness Value
	Scenario
	Figure 173: Space Utilization Datastore Cluster Prior to Initial Placement
	Figure 174: Datastore 1 Simulation Prerequisites Migrations
	Figure 175: Datastore 2 Simulation Prerequisite Migrations
	Figure 176: Datastore 3 Simulation Prerequisite Migrations
		Adding a New Disk to an Existing VM in a Datastore Cluster
		Manually Migrating VMs within the Datastore Cluster
	Figure 177: Datastore Cluster as Destination
	Figure 178: Disable Storage DRS Reveals Individual Datastores
	STORAGE DRS LOAD BALANCING
	Figure 179: Storage DRS Thresholds
		Space Load Balancing
	Figure 180: Space Load Balance Workflow
		Collecting Statistics
	Figure 181: Minimum Space Utilization Difference
		Cost Benefit-Risk Analysis
		Migration candidate selection
		On-Demand Space Load Balancing
		I/O Load Balancing
	Figure 182: I/O Load Balancing Workflow
		Stats Collection – Performance Snapshot
		Online Device and Workload Modelling
		Device Modelling
		Workload Modelling
		Normalized Load
	Figure 183: I/O Load-Balancing Input
		Data Points
	Figure 184: Invocation Period Overview
		Load Imbalance Recommendations
	Figure 185: I/O Imbalance Threshold
		Cost-Benefit Analysis
		Ignoring Peak Moments
		SIOC Latency and Storage DRS Latency
		Datastore Correlation Detector
		Load Balancing Recommendations
			Unified Recommendations
			Dependent Migration Recommendations
			Cultivation Time
		Invocation Triggers
			Cluster Configuration Change
			Datastore Maintenance Mode
			Initial Placement
			Exceeding Threshold
			Invocation Frozen Zone
		Recommendation Calculation
			How to Create a "New Storage DRS recommendation generated" Alarm
	Name and Description
	Figure 186: Alarm Name and Description
	Targets
	Figure 187: Alarm Targets
	Alarm Rule
	Figure 188: Alarm Rule
	Review
	DATASTORE CLUSTER CONFIGURATION
	Figure 189: Datastore Cluster Ecosystem Architecture
		Creating a Datastore Cluster
			Configuration Workflow
			Name and Location
	Figure 190: Datastore Cluster Name and Location
		Storage DRS Automation
	No Automation (Manual Mode)
	Fully Automated
	Figure 191: Storage DRS Automation
	Figure 192: Customized Storage DRS Automation Settings
		Storage Run Runtime Settings
	Figure 193: Storage DRS Runtime Settings
		Select Clusters and Hosts
	Figure 194: Cluster and Host Selection Screen
		Select Datastores
	Figure 195: Select Datastores
		Ready to Complete
	Figure 196: Ready to Complete
	ARCHITECTURE AND DESIGN OF DATASTORE CLUSTERS
		Connectivity
			Host Connectivity
			Compatibility List
	Figure 197: VM and VMDK Mobility in a Partially Connected Architecture
		I/O load balancing in Partially Connected Datastore Clusters
		Partially Connected Datastores and the Invocation Period
		Cluster Connectivity
	Figure 198: DRS and Storage DRS Load Balancing Domains
		Multiple Compute Clusters and SIOC
		Maximum Number of Hosts per Volume
		Array Connectivity
	Figure 199: Array Connectivity
		APIs for Storage Awareness
		Hardware Offloading
	Figure 200: VAAI Hardware Offloading within Arrays
		Datastores
			Space Utilization Threshold and the Space Safety Buffer
			Scale Up or Scale Out Datastores
		VM Configuration
			Datastore Cluster Default Affinity Rule
	Figure 201: Default VM Affinity
	Figure 202: Initial Placement with Default Affinity Rule
		DrmDisk
	Figure 203: DrmDisk of a VM
	Figure 204: Datastore Recommendation with Affinity Rules Disabled
		Increasing Granularity
	Figure 205: Initial Placement with VMDK Anti-Affinity Rule Enabled
		I/O Load Balancing with Anti-Affinity Rules
		Disk Types
			Thick Disk
			Thin Provisioned Disk
			Independent Disk
			Avoiding VMDK Level Over-Commitment While Using Thin Disks and Storage DRS
	IdleMB
	Figure 206: Thin Provisioned Disk
	PercentIdleMBinSpaceDemand
	Entitled Space Use
	Figure 207: Entitled Space Use
	Calculation During Placement
	Figure 208: Datastore Free Space
	Changing the PercentIdleMBinSpaceDemand Default Setting
	Use case 1: NFS Datastores
	Use case 2: Safeguard to Protect Against Unintentional Use of Thin Disks
		VM Automation Level
	Impact of VM Automation Level on Load Balancing Calculation
		Interoperability
			Array Features
			Array-Based Auto-Tiering
			Array-Based Deduplication
	Figure 209: VMDK Migration within Deduplication Pools
		Array-Based Replication
	Figure 210: VMDK Migration within Same Consistency Group
		Array-Based Thin-Provisioning
	Figure 211: VMDK Migration between Thin-Provisioning Pools
		Storage DRS Integration with Storage Profiles
			Profile-Driven Storage
	Figure 212: Compatible Datastore Cluster
	Figure 213: Multiple Storage Policies in Single Datastore Cluster
	Figure 214: Incompatible Datastores
	Figure 215: Datastore Selection Recommendations
	Figure 216: Compatible Datastore List
		Storage Policy Compliancy
	Figure 217: VM Storage Policy Compliance Status
	Prerequisites
	AFFINITY RULES
	Figure 218: Default VM Affinity
		Storage DRS Rules
	Figure 219: Storage DRS Rules
		VMDK Affinity Rule
	Figure 220: Initial Placement with Default Affinity Rule
		VMDK Anti-Affinity Rule
	Figure 221: Initial Placement with VMDK Anti-Affinity Rule Enabled
	Figure 222: Datastore Recommendation with Default Anti-Affinity
		VM Anti-Affinity Rule
	Figure 223: VM to VM Anti-Affinity Rule
	Figure 224: Default VM Affinity Conflict
		Violate Anti-Affinity Rules
			Anti-Affinity Rules with Datastore Correlation
	Figure 225: EnforceCorrelationForAffinity Enabled
		Overriding Datastore Cluster Default
		VM Overrides
	Figure 226: VM Overrides
		Storage DRS Rules
		Moving a VM into a Datastore Cluster
	DATASTORE MAINTENANCE MODE
		Automation Mode
	Figure 227: VM Evacuation Automation Level
	Using Datastore Maintenance Mode for Migration Purposes
		Datastore Maintenance Mode on a Datastore
		Throttle the Number of Storage vMotion Operations
	Figure 228: Storage vMotion in Progress
		How Do You Throttle The Number of Storage vMotion Operations?
	P4
	QUALITY CONTROL
	STORAGE I/O CONTROL
	Figure 229: Basic Shared Storage Architecture
	Figure 230: Unbalanced Storage I/O Consumption
	Figure 231: Balanced Storage I/O Consumption
		SIOC Explained
		Storage Fairness
			SIOC Defaults
	Figure 232: Default SIOC Configuration Values
		Latency Threshold Computations
			Automatic or Manual Threshold
		I/O Injector
	Figure 234: Migrate Cold Segments by Auto-Tiering
		Queue Depth
	Figure 235: Storage Path SAN
	Figure 236: Esxtop Disk Device View
		SIOC Logging
		Communication Mechanism
	Figure 239: Iormstats.sf Listing
	Figure 240: ESXi Host Accessing IORMSTATS.SF
		Local Scheduler
	Figure 241: Local Disk Scheduler
		Datastore-Wide Scheduler
	Figure 243: SIOC Share Variance Example
		VAIO
	Figure 244: VAIO Framework for SIOC
	Figure 246: Default SIOC Policy Components
	Figure 247: Custom SIOC Policy Component
	Figure 250: VM Storage Policy Compliancy
		Statistics Collection Only
	Figure 251: Enable Stats-Only Mode
	Figure 252: SIOC Performance Views
	Figure 253: SIOC Stats-Only Exemplary Performance View
		Storage I/O Allocation
		Shares
		Limits and Reservations
		Interoperability
	NETWORK I/O CONTROL
		Network I/O Control Constructs
	Figure 255: Basic Network I/O Control Constructs
		Evolution of NIOC
	Figure 256: Distributed vSwitch Versions in vSphere 6.7
	Figure 257: Verify NIOC Version in the GUI
		NIOC Defaults
	Figure 258: Default Configured NIOC Traffic Types in vSphere 6.7
	NIOC Advanced Setting
	Figure 259: Example NIOC Exclude Configuration
		Bursty Network Consumers
		Bandwidth Allocation
		Shares
	Figure 260: NIOC Shares Example
	Figure 261: NIOC Shares Distribution
	Figure 262: NIOC Shares Deviation with Fewer Traffic Sources
		Ingress and Egress Perspective
	Figure 263: NIOC Ingress and Egress Perspective
		Limits
	Figure 264: NIOC VDP Traffic Type Limit Configuration Example
	Figure 265: NIOC VDP Traffic Type Limit Example
		Test Scenario
	Figure 266: NIOC Traffic Type 3 Gbit/s Limit Test
	Figure 267: NIOC Traffic Type 2 Gbit/s Limit Test
		Scheduled Limit Values
		Destination Traffic Saturation
	Figure 268: NIOC Limit Consumption by Other ESXi Hosts
		Traffic Shaping
	Figure 269: Egress Traffic Shaping Configuration
	Figure 270: NIOC Limit Consumption Solved with Traffic-Shaping
		Reservations
	Figure 271: NIOC Maximum Reservation Value
	Figure 272: NIOC Reservation Example
		Network Resource Pools
	Figure 273: Network Resource Pool Constructs
	Figure 274: Virtual Machine Traffic Reservation Example
	Figure 275: Network Resource Pools
	Figure 276: Network Resource Pool on a Distributed Port Group
	Figure 277: Network Resource Pool Usage Overview
		Individual VM Parameters
	Figure 280: Power On Failure by vSphere DRS
		Bandwidth Management
		Traffic Marking
			Quality of Service
	Figure 281: QoS Classification Advertised from Virtual to Physical Layer
	Figure 282: CoS Tag Traffic iSCSI Rule Example
	P5
	STRETCHED CLUSTERS
	USE CASE: STRETCHED CLUSTERS
		Scenario
		Technical Requirements and Constraints
		Uniform Versus Non-Uniform vMSC Configurations
	Figure 283: Uniform Access
	Figure 284: Non-Uniform Access
		Infrastructure Architecture
			Infrastructure
	Figure 285: Our Lab Infrastructure
		vSphere Configuration
		vSphere HA
			Admission Control
	Figure 286: Cluster Creation
	Figure 287: Admission Control
		HA Heartbeats
	Figure 288: Heartbeat Datastores
	Figure 289: Configuration of Heartbeat Datastores
		Permanent Device Loss and All Paths Down Scenarios
	Figure 290: Failure Condition and Responses Configuration
	Figure 291: APD Response Delay
		Restart Ordering and Priority
	Figure 292: VM Group – Select a VM
	Figure 293: Restart Priority
	Figure 294: Start Next Priority VMs
	Figure 295: VM Group
	Figure 296: Restart Dependency
		ProActive HA
	Figure 297: Proactive HA Automated Remediation
	Figure 298: Proactive HA Provider
		DRS
			Site Affinity
	Figure 299: Former Required HA Rule Settings
	Figure 300: VM/Host Group Creation Part 1
	Figure 301: VM/Host Group Creation Part 2
	Figure 302: VM/Host Group Creation Part 3
		Advanced Settings
	Figure 303: DRS Additional Options
		Correcting Affinity Rule Violation
		Storage DRS
	Figure 304: Storage DRS Configuration
	Figure 305: Storage DRS Runtime Settings
		Migrations
		Failure Scenarios
			Single-Host Failure in Amsterdam Data Center
	Figure 306: Single Host Failure
	Result
	Explanation
		Single-Host Isolation in Amsterdam Data Center
	Figure 307: Single Host Isolation
	Result
	Explanation
		Storage Partition
	Figure 308: Storage Partition
	Result
	VMs remain running with no impact.
	Explanation
		Data Center Partition
	Figure 309: Data Center Partition
	Result
	VMs remain running with no impact.
	Explanation
	Figure 311: Lock Lost Message
		Disk Shelf Failure in Amsterdam Data Center
	Figure 312: Disk Shelf Failure
	Result
	VMs remain running with no impact.
	Explanation
		Full Storage Failure in Amsterdam Data Center
	Figure 313: Full Storage Failure Site Amsterdam
	Result
	VMs remain running with no impact.
	Explanation
		Permanent Device Loss
	Figure 314: Permanent Device Loss
	Result
	Explanation
	Figure 315: HA PDL Response Configuration
		Full Compute Failure in Amsterdam Data Center
	Figure 316: Full Compute Failure
	Result
	All VMs are successfully restarted in Rotterdam data center.
	Explanation
		Loss of Amsterdam Data Center
	Figure 317: Full Loss of Data Center
	Result
	All VMs were successfully restarted in Rotterdam data center.
	Explanation
		Summary
	INDEX
	A
	Adapter Queue
	B
	C
	Class of Service
	D
	DSCP 502
	DPM 114, 159, 173, 307
	E
	F
	Failures To Tolerate
	Fault Domain Manager
	Fault Tolerance
	H
	High Availability
	M
	Microsoft Clustering Services
	N
	Network File System
	Network I/O Control
	Network Resource Pools
	O
	Outstanding I/O
	P
	Permanent Device Loss
	Q
	Quality of Service
	R
	Raw Device Mappings
	Return On Investment
	S
	Start-time Fair Queuing
	Storage Device
	Storage I/O Control
	T
	THLSD 262, 263, 271
	V
	Virtual Machine
	VMDK 330
	VMFS 445
	Virtual SAN
	Virtual Volumes
	VM Component Protection
	HCL 472
	VVD 476
	VAAI 394
	VAIO 445, 463
	VASA 102, 394, 407
	VAMI 183
	VDP 480, 486
	W
	WSFC 19