By Lisa Gecko
VMware’s just released vSphere 5 is the latest upgrade to the industry’s most comprehensive suite of infrastructure virtualization products. With this release, VMware has upped the ante with new features designed to nearly eliminate downtime from the corporate network infrastructure. Some of these enhancements include improved vMotion, Storage vMotion, vSphere High Availability (HA), and VMware Fault Tolerance (FT).
vMotion is VMware’s technology for hot-migration of a virtual machine (VM) from one host to another. In a highly dynamic data center, physical servers require both software and hardware maintenance, both of which require the host to be shut down. With vMotion, VMs can be seamlessly moved from one host to another, enabling the maintenance of the underlying hardware without disruption. In previous versions, if a server was under high memory load, it can prevent a vMotion to proceed successfully. In this new version, VMware has implemented Slowdown During Page Send (SDPS), which can slow down the VM for a short period of time to enable the successful completion of a vMotion. A new feature called Metro vMotion has been added. This feature increases the maximum allowable latency for vMotion to 10ms. This enables organizations to move VMs over larger distances without disruption.
Improved Storage vMotion
Like vMotion, Storage vMotion allows a VM to be moved from one storage medium to another without disruption, enabling SAN maintenance and SAN load balancing without the need for downtime. VMware has enabled migration between hosts with dissimilar storage, i.e. from Direct Attached Storage to a SAN. Storage vMotion performance has also been enhanced with I/O mirroring in vSphere 5.
Improved VMware High Availability
VMware High Avalailability (HA) is a host clustering technology that protects VMs from underlying hardware failure. In the event of a host failure, VMs on a host that becomes unavailable will be immediately started on another host in the HA cluster. Additionally, if a VM becomes unresponsive, HA can automatically restart the VM. This enables automatic recovery of a system before a system administrator becomes aware of the problem. In version 5, VMware has increased the number of maximum hosts in a cluster.
vSphere Fault Tolerance (FT)
vSphere Fault Tolerance takes HA one step further. FT is enabled on a VM by VM basis. When enabled, vCenter immediately creates a second VM that shares the underlying virtual disks. If the primary VM stops communicating, the secondary VM takes over. This is enabled by using vLockstep technology. vLockstep mirrors the CPU and memory state of the primary VM to the secondary in real-time. If a failure occurs, the affected VM is automatically recreated. This configuration enables recovery with zero downtime.
Over time it is becoming clear that highly available workloads should be virtualized. Early adopters who virtualized non-production servers and servers on the fringe are now virtual nearly everywhere. For all but the most intense workloads, virtualization enables organizations to protect their systems with little change to their applications. As VMware continues to improve their products, there will be fewer and fewer situations where virtualization will not be the norm.