It has been quite a while since I wrote as I was busy working on few projects, So lets continue to discuss advanced features in vSphere and also Upgrading scenarios , Monitoring and Troubleshooting vSphere using both GUI and command line interface. Firstly lets discuss some advanced features :
VMware High Availability
- This is an advanced cluster level operation which quickly brings back the mission critical business applications in the event of an ESXi Host failure.
- VMHA works on the concept of Master/Slave that is all clustered Hosts are contested in polling and a Master Host is elected
- VMHA installs an agent called Fault Domain Manager (FDMS) in every clustered Host that talks to vpxd (vCSA mgmt agent)
- FDMS is responsible for monitoring the state of the Slave Hosts and also the guest OS failures
- VM’s that are HA protected communicates to FDMS using the VMware tools installed on the guest OS
- Decreases downtime and improves availability for mission critical business applications (Web, Exchange, SQL etc)
- It works on the basis of VMware vMotion and must satisfy all requirements for vMotion
- VMHA monitors through both Management and Datastore heartbeat
VMware Fault Tolerance
- Another advanced vSphere VM level feature which provides Super high availability or continues availability for VM’s
- When a VM is put under FT it has zero downtime and shadow copy of the VM is created on a different Host
- VMware uses vLockstep technology where every writes on VM are replicated to shadow copy of VM on a different Host
- When a Hosts fails, VM running under FT is already up and running on a different Host and it is made primary VM
- FT also starts creating new shadow copy VM on clustered Host
- We must remember that VM hard disk or vmdk is never moved and resides on a shared storage
- We can install Site Survey plugin for vSphere client to check if our vSphere infra has all prerequisites for FT
Distributed Resource Scheduler
- DRS is an advanced cluster level operation which understands the resources and consumption of the virtual infrastructure
- Compute resources both CPU and Memory are monitored by DRS and Power is monitored using DPM
- DRS therefore promotes efficient use of physical resources and prevents any one VM to consume all resources
- DRS has 3 modes of operation that is Manual, Automated and Fully Automated modes
- DRS functionality depends on what mode we choose VMware recommends Fully Automated mode
- DRS also provides us affinity/anti affinity rules for placements of VM’s on desired Hosts
Storage Distributed Resource Scheduler
- VMware vSphere 5 onwards SDRS was introduced and is a revolutionary technology
- SDRS is an advanced datastore cluster level operation where datastores are pooled together to create datastore cluster
- Traditional provisioning of VM includes manual selection of Datastore
- There is no actual guarantee for the performance of the VM due to datastore latency or number of VM’s on datastore
- SDRS allows VM’s to be placed on a datastore with least latency and most space available
- VM Hard disks or vmdk can be migrated to a different datastore if there is a performance issues due to storage
- SDRS also provides us affinity/anti affinity rules for placements of VM’s on desired datastore
Network I/O Control
- NIOC was introduced from VMware vSphere 4.1 onwards
- NIOC provides the quality of service for network traffic in the vSphere infrastructure
- NIOC uses different types of traffic (vMotion, VM Traffic , VMkernel Traffic) and prioritizes them
- NIOC works at distributed switch level (vDS) and hence only with a enterprise plus license
- NIOC basically uses Shares, Limits and Reservations on network bandwidhts for VM’s
- NIOC ensures best performance for mission critical vSphere infrastructure services
Storage I/O Control
- SIOC was introduced from VMware vSphere 5 onwards
- SIOC also provides the quality of service to ensure that business critical VM’s get storage performance they need
- SIOC helps us to identify mission critical VM’s and are not slowed down by lower priority VM’s
- SIOC is enabled per datastore, and ESXi will monitor the datastore latency and performance
- SIOC also uses Shares, Limits and Reservations for individual Hosts or VM’s
Auto Deploy
- Auto deploy was introduced from vSphere 5 onwards
- Auto deploy allows us to quickly deploy ESXi on more than stateless physical Hosts
- While creating Auto deploy we must also create Rule Set
- Rule Set provides set of criteria’s for physical Hosts to deploy the correct ESXi images on appropriate Hosts
- With Auto deploy we have set of pre provisioned Hosts that are attached to network
- Auto deploy would then detect OEM’s of Host and install appropriate ESXi and drivers on them
How does it work ?
- Stateless Host PXE boots and gets an IP address from DHCP
- DHCP server will then point the Host to TFTP server
- We must configure DHCP and option 66 to the TFTP server and option 67 to configuration file called gPXE
- This file gPXE gets downloaded on to Host
- gPXE configuration file instructs the Host to make Http boot requests to auto deploy
- Auto deploy queries the Rule Set engines to get more information about physical Host
- An image profile and a Host profile is attached to the physical Host based on Rule Set
- ESXi is installed to RAM and is added to vCenter and configured
- vCenter maintains image and Host profiles for each Host in the database