20160129-VMware-VSOS-Day 5

20160129-VMware-VSOS-Day 5

Module 11 Virtual Machine and Cluster Optimization

Lesson 1: Virtual Machine Optimization

Storage Considerations

• VMware Paravirtual SCSI ( PVSCSI )
• http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=1010398
• 效能會比 LSI Logic and LSI SATA較好, ESXi 5.0 以上都支援
• 可以參考 http://longwhiteclouds.com/2014/01/13/vmware-vsphere-5-5-virtual-storage-adapter-performance/

Notes:

• VM開機時, 顯示被 Lock 無法開機, 刪除 *.LCK 可以解決

Lesson 2: vSphere Cluster Optimization

-- Class break --

Notes:

• 清除 vSphere Client 連線 IP
• http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=1032988
• 把 HKEY_CURRENT_USER\Software\VMware\VMware Infrastructure Client\Preferences 內的 RecentConnections 的 IP 清除

Lab 12: Diagnosing vSphere HA Cluster Resource Issue

• 參考 Lab 文件
• Task 1: Shutdown all running virtual machine
• Task 2: (Student B) Disconnect the ESXi Host from the vCenter Server Instance
• Task 3: (Student A) Connect a Second ESXi Host to the vCenter Server Instance
• Task 4: Review Change in the vSphere Web Client
• Task 5: Move your ESXi Host to the First Problem Cluster
• Task 6: Investigate the First Cluster Issue
• Task 7: Move Your ESXi Host to the Second Problem Cluster
• Task 8: Investigate the Second Cluster Issue and Start your first Virtual Machine
• Task 9: (Student A) Start a second virtual machine in the second cluster
• Task 10: (Student B) STart a second virtual machine in the second cluster
• Task 11: Determine the cause of the second cluster issue
• Task 12: Move Your ESXi Host Back to the training data center
• Task 13: (Student B) Clean up for next lab

-- 中午休息 --

Module 12 Host and Management Scalability

Lesson 1: Host Profiles

備份 Host Profile

• 透過 vCLI 執行 vicfg-cfgbackup --server IP或FQDN -s  檔案名稱 來備份
• 還原使用 vicfg-cfgbackup --server IP或FQDN -l  檔案名稱

Host Profile用途

• 備份 / 還原主機設定
• 套用到多台主機
• 用 Host profile 檢查主機現有的設定

Lab 13: Host Profiles

• 參考 Lab 文件
• Task 1: Create and Export a Host Profile
• 沒有加入主機的 Host Profile 就是包含網路,storage與驗證來源的設定
• Task 2: Import a Host Profile
• Task 3: Attach an ESXi Host to the Imported Host Profile
• Task 4: Run an initial compliance check
• Task 5: Introduce a configuration drift
• Task 6: Run a compliance check and remediate the configuration drift
• Task 7: Introduce a second configuration drift
• Task 8: Run a compatibility check and try to remediate the host
• Task 9: Detach the host profile and exit maintenance mode

Lesson 3: vSphere PowerCLI

• 可以參考 https://www.vmware.com/support/developer/PowerCLI/PowerCLI55/html/

Lab 14: Using vSphere PowerCLI

• 參考 Lab 文件
• Task 1: Start vSphere PowerCLI
• Task 2: Define Variables
• Task 3: Connect to and Disconnect from an ESXi Host
• Task 4: Connect to the vCenter Server System and View ESXi Host-Related Information
• Task 5: List Virtual Networking Properties
• Task 6: Retrieve Virtual Machine Properties
• Task 7: Retrieve Virtual Machine Performance Statics

-- Class break --

Lesson 4: vCenter Converter and vSphere ESXi Image Builder

ESXi Images

• An ESXi image is a software bundle that consists 4 main components
• Core Hypervisor
• CIM Providers
• Plug-In Components
• Drivers

Lesson 5: vSphere Auto Deploy

-- Class End --

20160128-VMware-VSOS-Day 4

20160128-VMware-VSOS-Day 4

Lab 10: Monitoring CPU Performance

• 參考 Lab 文件,  做 Task 7 ~ Task 12
• Task 7: Run a Single-Threaded Program on a Virtual Machine with Two vCPUs
• Task 8: Record Statistics for Case 2: ONe Thread and Two vCPUs
• Task 9: Run a Dual-Threaded Program in a Dual-vCPU Virtual Machine
• Task 10: Record Statistics for Case 3: Two Threads and Two vCPUs
• Task 11: Analyze the Test Results
• Task 12: Clean Up for the Next Lab

Lesson 3: Troubleshooting CPU Performance Problems

Host CPU Saturation

• Ready Time
• Web client 測量 > 2000 ms 代表 CPU 不夠
• resxtop 測量 > 10% 代表 CPU 不夠

判斷 CPU 方式

• 是否大於 2000ms 或是 10%, 超過列出清單
• USE time 如果與 Ready Time 差不多則視為正常

Kernel 是單CPU

• 有可能自行更換 kernel
• msconfig 限制

-- Class break --

Module 8 Storage Optimization

Lesson 1: Storage Virtualization Concepts

Storage Protocol Performance

Fibre Channel

• 16 Gbps
• 8 Gpbs
• 4 Gpbs

FCoE

• 10 Gbps

Software and hardware iSCSI and NFS support for jumbo frames

• Using Gigabit, 10 Gb, and 40 Gb Ethernet NICs
• Using 10 Gb iSCSI hardware initiators

Storage Queues

• Storage queue types
• Queueing at the host
• VMkernel queue
• Device Driver queue
• Maximum recommended queue depth is 64( default for VMware, other brands depth default 32 )
• 調整方式 http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=1267
• Queueing at the storage array
• Storage Array queue
• 透過 resxtop 判斷

Lesson 2: Monitoring Storage Activity

About Disk Capacity Metrics

• 觀察 Latency 與 aborted disk command

Disk Throughput Example

IOPS = READs / s + WRITEs / s

GAVG = DAVG + KAVG

• GAVG
• VM的 Disk latency
• DAVG
• Device 的 latency
• KAVG
• VMKernel 的 latency

Disk Latency

• resxtop 上面的 QUED 為 queue 數量

Lab 9 Monitoring Storage Performance

• 參考 Lab 文件
• Task 1: Prepare the Test Virtual Machine
• Task 2: Prepare the IP Storage Network for Testing
• Task 3: Create a Real-Time Disk I/O Performance Chart
• Task 4: Prepare to Run Tests
• Task 5: Measure Continuous Sequential Write Activity to a Virtual Disk on a Remote Datastore
• Task 6: Measure Continuous Random Write Activity to a Virtual Disk on a Remote Datastore
• Task 7: Measure Continuous Random Read Activity to a Virtual Disk on a Remote Datastore
• Task 8: Measure Continuous Random Read Activity to a Virtual Disk on a Local Datastore
• Task 9: Analyze Test Results
• Task 10: Clean Up for the Next Lab

-- 中午休息 --

Lesson 3: Troubleshooting Storage Performance Problems

Overloaded Storage and Command Aborts

• If commands aborted (ABRTs/s) > 0  for any LUN, then storage is overloaded on that LUN

Slow Storage

You can also monitor device latency (DAVG) in resxtop

• If value >10, a problem might exit.
• If value > 20, a problem exists. ( Maybe due to disable Cache )

Module 10 Memory Optimization

Lesson 1: Memory Virtualization Concepts

Virtual Machine Memory-Reclamation Techniques

• Transparent page sharing
• Disabled by default with ESXi 6.0, ESXi 5.5 update 2 and later set disable
• VM 共享記憶體
• Guest OS 版本相同

如果記憶體不充足

• TPS
• Ballooning
• Memory Compression
• Host Swap

Memory Overhead

• 根據記憶體大小以及 vCPU 數量決定 VMM 大小

Lesson 2: Monitoring Memory Activity

Monitoring Virtual Machine Memory Usage

Notes:

• Memory 相關名詞
• vSphere Client
• Consumed 消耗記憶體
• Active 正在使用記憶體
• Consumed - Active = 閒置記憶體
• vSphere Web Client
• VM overhead consumed - VMM
• Utilization / Active - 配置/正在使用中記憶體

Using resxtop to Monitor Memory Usage ( 好用資訊 )

• 透過 resxtop 的 state 來觀察記憶體狀況以及會啟用哪些機制
• 按 m 觀察記憶體
• High / Soft / Hard / Low state 會採取不同機制
• PMEM - 實體記憶體
• VMKMEM - VMkernel 記憶體
• PSHARE - TPS 記憶體
• SWAP - Host Swap 記憶體
• ZIP - Memory Compression 記憶體壓縮
• MEMCTL - Ballooning 記憶體

-- Class break --

Lab 11: Monitoring Memory Performance

• 參考 Lab 文件
• Task 1: Generate Database Activity in the Test Virtual Machine
• Task 2: Check for Overcommittment of Virtual Machine Memory
• Task 3: Configure an Advanced Performance Chart
• Task 4: Configure resxtop to Report Virtual Machine Memory Statistics
• Task 5: Observe Baseline Memory Statistics
• Task 6: Import the RAM-HOG vApp
• Task 7: Start a Memory Test on Each RAM-HOG Workload
• Task 8: Record Memory Statistics

Lesson 3: Troubleshooting Memory Performance Problems

VMware 啟用 TPS 參考  

• http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=2097593

-- Class End --