Module 3: vCenter Scalability Features and Benefits
Welcome to module 3, vSphere Scalability Features and Benefits. These are the topics that will be covered in this module.
Student Study Guide – VTSP 5.5
Module Objectives
In this module we are going to take a closer look at the distributed services that vCenter manages and enables. These services provide the cluster wide features and advanced functionality that are the key to vSphere scalability.
Many aspects of this module focus on not only explaining the distributed services but also showing you sample whiteboards of how to present them to customers.
If a particular service/feature/functionality is not a part of the core vSphere Standard licenses we will mention which license tier in which the feature is available.
VTSP 5.5 - Student Study Guide
Presenting vMotion
The traditional challenge for IT is how to execute operational and maintenance tasks without disruption to business service delivery. In a non-virtualized environment this essentially means downtime whenever maintenance is required on the infrastructure.
In a virtualized environment we have virtual machines running on the hypervisor which is installed on the physical host.
Using vMotion, we can migrate (move) a virtual machine from one host to another host without incurring any downtime. To do this we have to copy the memory footprint and the running state of the virtual machine progressively across to a separate physical host and then switch over to the running instance of the virtual machine to the new host, all without any loss of data or downtime.
This may require a significant amount of dedicated network bandwidth as we may be moving a lot of data between each host. Sufficient network bandwidth should be considered during the planning, implementation and configuration stages (On-going
Student Study Guide – VTSP 5.5
Performing a vMotion without shared storage is possible using Enhanced vMotion.
However, this will require a significantly longer time to copy a virtual machine as all the VM file data AND memory footprint has to be copied across a network before the virtual machine can be completely moved to the destination host.
A benefit of using shared storage is that vMotion only needs to move the memory footprint and CPU state of the running virtual machine from the memory of one physical host to another. vMotion with shared storage therefore completes in less time than vMotion that does not use shared storage. Prior to version 5.1, this was the only option for live migration of VMs between hosts. The term vMotion on its own typically implies that shared storage is required.
While vMotion allows us to migrate virtual machines between hosts so that we can carry out maintenance operations on those hosts without any downtime, then how do we do the same for storage? Storage vMotion allows us to move the virtual machine files between storage systems that are connected to the same host, without any service interruption for the users.
All of these vMotion capabilities enable you to load balance between physical hosts and storage systems, carry out planned hardware maintenance without service
interruption and they also allow you to carry out system expansions or upgrades such as moving to newer faster hosts or storage systems without incurring service
interruptions.
For vMotion to be supported you require the following:
vMotion requires as a minimum the Essentials Plus license.
Virtual machine files are located on shared storage. Shared storage can be on a Fibre Channel or iSCSI storage area network (SAN), or an NFS NAS. Enhanced vMotion does not require shared storage but all other pre-requisites are the same as for standard vMotion. Hosts must have the same CPU manufacturer and family or have the same Enhanced vMotion Compatibility (EVC) baseline.
Migration with vMotion requires correctly configured network interfaces on source and target hosts. On each host, configure a VMkernel port group for vMotion. It is a best practice to have at least one dedicated GigE adapter for vMotion per host.
You must ensure that virtual machines have access to the same subnets on source and destination hosts. If you are using standard switches for networking, ensure that the network labels used for virtual machine port groups are
consistent across hosts.
If you are using vSphere Distributed Switches for networking, ensure that source and destination hosts are members of all vSphere Distributed Switches that virtual machines use for networking.
Use of Jumbo Frames is recommended for best vMotion performance.
VTSP 5.5 - Student Study Guide
Presenting HA
We‟ve talked about planned outages and how to manage them using vMotion. What happens if you have an unplanned outage and how do we deal with it?
In the case of multiple ESXi hosts with virtual machines running on them, and where files used shared storage, we can restart virtual machines on other available hosts if one of our hosts fails.
This technology is called vSphere High Availability or vSphere HA.
vSphere HA provides high availability for virtual machines and the applications running within them, by pooling the ESXi hosts they reside on into a cluster.
Hosts in the cluster are continuously monitored. In the event of a host failure, the virtual machines on the failed host attempt to restart on alternate hosts.
vSphere App HA is new in vSphere 5.5. It is a virtual appliance that you can deploy on the vCenter Server.
Student Study Guide – VTSP 5.5
In designing the system, you must ensure that you have enough capacity for HA recovery from a host failure. HA performs failover and restarts virtual machines on different hosts. Its first priority is the immediate availability of all virtual machines.
If you have hosts with too many virtual machines and you don‟t have enough capacity, some of the virtual machines might not start even if HA is enabled.
However, if you do not have sufficient resources, you can prioritize your restart order for the most important virtual machines.
Virtual machines are restarted even if insufficient resources exist, but you now have a performance issue because virtual machines contend for the limited resources.
If virtual machines have reservations and those reservations cannot be guaranteed, then some virtual machines might not be restarted.
You should also implement redundant heartbeat network addresses and isolation addresses, and address the possible issue of a host isolation response (in the case where a master heartbeat is lost).
As a minimum, you require an Essentials Plus License to use HA. vSphere App HA is only available with an Enterprise Plus License.
VTSP 5.5 - Student Study Guide
Presenting DRS
A major concern in IT environments is to ensure that the load is distributed effectively across the available resources. In this example we have two hosts handling all of the load and one host with no load. Ideally, you want your infrastructure to sense this and to move the loads so that the overall utilization is balanced.
VMware DRS or distributed load balancing monitors host, CPU and memory utilization and can automatically respond to changes in load by rebalancing virtual machines across the cluster when necessary. These virtual machines will be moved using vMotion.
As new virtual machines are created or started, DRS can decide the optimal placement for the virtual machines so that CPU and Memory resources are evenly consumed across the cluster.
When you add a new physical server to a cluster, DRS enables virtual machines to immediately take advantage of the new resources because it re-distributes the running
Student Study Guide – VTSP 5.5
A Combination of HA and DRS can be used to enhance the clusters response to host failures by improving the load distribution of the restarted VMs. HA powers on VMs then DRS load balances VMs on Hosts.
Using DRS you can choose different levels of automation, Fully automated, Partially automated and Manual. Fully Automated automatically places virtual machines on hosts when powered on, as well as automatically migrating virtual machines. Partially automated means that virtual machines will be automatically placed on hosts when powered on and vCenter will suggest migrations. Manual will suggest migration recommendations for virtual machines.
When designing DRS it is important to ensure that you have shared storage. As DRS uses vMotion you should ensure that you understand the design requirements for vMotion setup.
As a minimum you require an Enterprise License to use DRS.
VTSP 5.5 - Student Study Guide
Presenting DPM
Dynamic Power Management is an enhancement to DRS which monitors overall
utilization across a cluster and if it finds that the required protection levels can be met by running all VM‟s on a reduced number of hosts, it will evacuate all virtual machines from one or more hosts and then put those hosts into standby mode in order to save overall power consumption.
When the virtual machine load increases and the host can no longer provide the required level of protection, DPM will automatically restart hosts and migrate virtual machines back onto them once they have come back online. When configuring DPM, you must ensure that the cluster can startup and shutdown each host using IMPI or Wake on LAN.
You should configure the vSphere DPM automation level for automatic operation and use the default vSphere DPM power threshold. This decreases power and cooling costs as well as decreasing administrative management overhead.
Student Study Guide – VTSP 5.5
Presenting FT
In some cases it is desirable to have the absolute minimum risk of downtime for some virtual machines. VSphere fault tolerance or FT maintains an identical copy of a running virtual machine in lockstep on a separate host. All Inputs and events performed on the primary virtual machine are recorded and replayed on the secondary virtual machine ensuring that the two remain in an identical state.
FT ensures that in the case of a host failure, the lockstep copy instantly takes over with zero downtime.
vSphere Fault Tolerance is currently limited to Virtual Machines with a single vCPU and, like vMotion, dedicated network uplinks on all hosts are recommended in order to
ensure that there is sufficient bandwidth available. All other standard vMotion
constraints also apply to VMs protected with Fault Tolerance. Consider that you are using twice the amount of resources, so factor this into your design when using FT.
As a minimum you require a Standard License in order to use FT.
VTSP 5.5 - Student Study Guide
Presenting Storage Distributed Resource Scheduler (SDRS)
Just as DRS is utilized to provide benefits such as resource aggregation, automated load balancing and bottleneck avoidance for host resources, Storage DRS provides the same capabilities for storage resources. You can group and manage a cluster of similar datastores as a single load-balanced storage resource called a datastore cluster.
Storage DRS collects the resource usage information for this datastore cluster and makes automated decisions or recommendations about the initial virtual machine file placement and migration to avoid I/O and space utilization bottlenecks on the
datastores in the cluster.
Storage DRS affinity rules enable controlling which virtual disks should or should not be placed on the same datastore within a datastore cluster. By default, a virtual machine's virtual disks are kept together on the same datastore. Storage DRS offers three types of affinity rules:
VMDK Anti-Affinity:
Student Study Guide – VTSP 5.5
VM Anti-Affinity:
Two specified virtual machines are kept on different hosts from each other.
You should use affinity and anti-affinity rules as needed. As an example, to improve the performance of an application by keeping the application disk on a datastore separate from the operating system disk.
As a minimum you require an Enterprise Plus license to use Storage DRS.
VTSP 5.5 - Student Study Guide
Presenting Host Profiles
A host profile captures the configuration of a specific host in a template form so that this profile can then be used to configure other hosts or validate if a host‟s configuration meets the requirements set by the administrator. This greatly reduces the manual steps involved in configuring hosts and maintaining consistency and correctness in host configuration across the data center.
Host profiles eliminate per-host, manual or UI-based host configuration. vSphere administrators can use host profile policies to maintain configuration consistency and correctness across the data center. Host profile policies capture the blueprint of a known, validated golden configuration and use this as a baseline to configure
networking, storage settings, security settings, and other settings on multiple hosts. This baseline can then be used to do a one-click or even scheduled configuration of newly-discovered or re-provisioned hosts.
vSphere administrators can also monitor changes to this baseline configuration, detect
Student Study Guide – VTSP 5.5
Presenting Storage Profiles
Profile-Driven Storage enables administrators to have greater control over their storage resources. It enables virtual machine storage provisioning to be automatically defined by the configuration of the virtual machine.
Profile-Driven Storage uses VASA to deliver the storage characterization supplied by the storage vendors to vCenter. VASA improves visibility into the physical storage infrastructure through vCenter Server. Storage can also be manually tagged by the Administrator. Instead of only seeing a block or file device with some amount of capacity, this allows vCenter to know about replication, RAID, compression, deduplication, and other capabilities and characteristics of the storage that it has available.
With this new information, VMware administrators can create storage profiles that allow them to define storage in terms of capability and not just capacity. Virtual machines can then be assigned storage by policy in order to meet their storage performance,
protection or other characteristics in addition to capacity. The storage characterizations are used to create the virtual machine placement rules in the form of storage profiles.
Storage Profiles also provide a way to check a virtual machine‟s compliance against these rules.
A key benefit here is that the virtual machine administrator no longer needs to make
VTSP 5.5 - Student Study Guide
machines. This also extends to templates so that rapid provisioning can also automatically assign suitable storage without requiring detailed storage allocation decisions.
For example machines may have their storage automatically selected based on a service level agreement which matches a particular storage profile. In the case shown here, the virtual machine has Gold level storage specified for both its configuration files and one of its virtual disks and, Bronze level storage for the other virtual disk. In this environment, the Gold Storage profile corresponds to a datastore that has fast disks with a high performance RAID type. The Bronze profile corresponds to a datastore that has slower disks with a space-efficient RAID type. The use of storage profiles makes it easier for the administrators to ensure that the correct types of storage resources are automatically selected.
If you are planning to use storage profiles ensure that your storage system has VMware vStorage APIs for Storage Awareness (VASA).
Because multiple system capabilities for a datastore are not supported, a datastore that spans several extents assumes the system capability of only one of its extents.
Storage profiles require as a minimum an Enterprise Plus License.
Student Study Guide – VTSP 5.5
Distributed Virtual Switches
In a vSphere infrastructure, if we need to make any changes or add capacity to the network configuration in the cluster, we have to ensure that the changes are configured correctly across each host, or we run the risk of breaking services such as vMotion, where incorrectly labeled port groups will cause vMotion to fail. Managing the network configuration on each host becomes a significant administrative overhead when dealing with multiple servers.
The vSphere Distributed Switch (vDS) improves virtual machine networking by enabling you to set up virtual machine access switching for your entire data center from a
centralized interface.
This enables the following , improved virtual machine network configuration, enhanced network monitoring and troubleshooting capabilities and support advanced vSphere networking features such as LACP.
The vSphere Distributed Switch provides the building blocks for many advanced
networking features in a vSphere environment, such as PV-LANs, VXLAN, Network I/O control, SR-IOV and third-party switch extensions.
vSphere 5.1 also added the ability to export and restore DV switch configurations which enhances the supportability of virtual switches. It also allows the use of the export as a template to create a distributed switch in any other deployment.
VTSP 5.5 - Student Study Guide
DV switches now also support an automatic configuration rollback. This ensures that misconfigurations that would significantly impact host or distributed switch behavior are prevented from taking effect by rolling back to the previous valid configuration. These rollbacks will occur when a network change that disconnects a host or creates any other invalid host networking configuration. Rollbacks will also occur when an invalid update is attempted on a distributed switch, distributed port group or distributed ports.
As a minimum you require an Enterprise Plus license to use Distributed Virtual Switches.
Distributed Virtual Switches are covered in more depth in Course 4.
Student Study Guide – VTSP 5.5
Auto Deploy
vSphere Auto Deploy facilitates the rapid provisioning of vSphere hosts by leveraging the network boot capabilities of x86 servers together with the small footprint of the ESXi hypervisor. Once installed a vCenter host profile is used to configure the host. After configuration the host is connected to vCenter where it is available to host virtual machines. The entire process is fully automated allowing new hosts to be quickly provisioned with no manual intervention.
Stateless or diskless caching host deployments let you continue operation if the Auto Deploy server becomes unreachable. This was the only mode of install available in vSphere 5.0.
Stateless Caching caches the image when you apply the host profile. When you later reboot, the host continues to use the Auto Deploy infrastructure to retrieve its image. If the Auto Deploy server is not available, the host uses the cached image.
VTSP 5.5 - Student Study Guide
vSphere Storage Appliances
A VSA cluster leverages the computing and local storage resources of several ESXi hosts. It provides a set of datastores that are accessible to all hosts within the cluster.
An ESXi host that runs VSA and participates in a VSA cluster is a VSA cluster member.
A VSA cluster is an affordable resilient shared storage solution that supports vMotion, HA, DRS and other vSphere distributed services that is easy to install, and expandable.
Multiple VSA instances can be managed within a single vCenter Environment.
With VSA 1.0, you could create a VSA cluster with two or maximum three VSA cluster members.
As of vSphere 5.1 the VSA can now support disks in all ESXi hosts in the cluster,
supports up to eight 3TB or twelve 2TB disks internally and up to 16 external disks in an expansion enclosure.
The VSA now has the ability to dynamically enlarge the shared storage managed by the VSA provided there is sufficient additional free physical space on all nodes in the VSA
The VSA now has the ability to dynamically enlarge the shared storage managed by the VSA provided there is sufficient additional free physical space on all nodes in the VSA