Sergey Nivens - Fotolia
VM monitoring tools are an essential part of creating and managing a virtual environment. VMware vSphere HA offers features such as disk and network monitoring, heartbeat customization and affinity rules, all of which are helpful for IT administrators who oversee virtual environments.
Once admins have set up vSphere High Availability (HA) correctly, they can use it to monitor and restart VMs. By using tools such as Distributed Resource Scheduler (DRS) and Admission Control in conjunction with vSphere HA, admins can address VM issues such as failed restarts and resource limitations.
Set up a vSphere HA cluster
Though vSphere HA offers different features to protect a virtualized environment, admins can't use it to its full potential without proper setup. VMware vSphere HA is a cluster technology that uses a Fault Tolerance feature, which prevents downtime for workloads that run on VMs and helps avoid a failure in a clustered environment.
To run vSphere HA, admins need at least two ESXi hosts managed by vCenter Server and a form of shared storage. Otherwise, they won't be able to protect VMs that run on a host's local storage. Also, admins must build redundancy into their network design by incorporating at least two network interface cards (NICs). By doing this, they can connect two NICs to different physical switches to improve management network dependability. This ensures that the cluster is more resilient and permits other vSphere HA capabilities, such as restarting VMs in the event of a failure, even if they reside on other hosts.
Monitor VMs with vSphere HA
VMware vSphere HA offers admins the ability to monitor their VMs to avoid unresponsiveness and unnecessary VM restarts. VM unresponsiveness often occurs when a VM's heartbeat is absent for longer than a predetermined time, at which point vSphere indicates either a fault in the guest OS or an underlying issue with VMware Tools.
VM Monitoring in vSphere HA can help check the VM's I/O to determine disk or network activity that might also account for unnecessary VM restarts. VMware vSphere HA also offers many other features to combat these issues -- such as application monitoring, application monitoring sensitivity settings and heartbeat customization -- which enable admins to detect a lack of compute time and manipulate a VM's heartbeat. These options enable quicker fault detection and earlier VM restarts, but sometimes false positives can occur.
Configure vSphere HA for VM restarts
VMware vSphere HA can restart VMs if it detects a failure as long as the system meets five criteria: To restart a VM, the cluster's master host must evaluate VM file accessibility, suitable host system availability, potential host resources, host limits and VM affinity or anti-affinity rules.
These predetermined constraints enable admins to eliminate host systems that can't support a VM restart and determine whether or not their selected system can support the VM's requirements.
Use DRS with vSphere HA
By supporting HA with various methods, admins can reduce the number of failed VM restarts due to resource limitations and overloaded VMs. Using DRS and HA together can address this issue by enhancing HA with affinity rules.
VMware vSphere HA paired with DRS can result in evenly loaded HA clusters by bringing hosts out of standby mode so vSphere HA can perform a failover. When using DRS affinity rules, admins must choose whether vSphere HA respects these rules or favors DRS VM anti-affinity limitations.
Fail over with vSphere Admission Control
VMware vSphere Admission Control is another tool for VM failover. In a VMware environment, failover depends on adequate resources in the specified host system. Admission Control can be an answer to this issue for admins, because the tool enables a target host system to reserve the resources required for a VM failover, which ensures the VM cluster node can fail over to the selected host.
To do this, admins must determine the amount of resources necessary for failover tasks with the use of vSphere HA and Admission Control. VMware vSphere HA and Admission Control use three approaches to reserve resources: setting aside a portion of the HA cluster's CPU and memory, using a slot policy to calculate and reserve resources and using dedicated failover hosts in the cluster.