In the last few years VMware has added a number of features to its core vSphere platform to address organizational concerns about the availability and uptime of their virtualized applications, to include High Availability (HA), vCenter Site Recovery Manager (SRM) and vMotion. Yet there are still other aspects of delivering on the uptime requirements of mission critical applications that enterprises want that VMware does not offer. It is these gaps that Symantec’s upcoming release of Veritas Cluster Server fills.
Few enterprises would argue that the investments they make in keeping their virtualized mission critical applications up and operational are well worth it to avoid whatever the real or perceived costs might be if they experience any type of outage. However the real costs alone are likely to prompt most enterprise organizations to act sooner rather than later.
Results in the recently released September 2012 State of the Data Center survey revealed that the average enterprise experienced 16 outages in the data center a year that added about $5.1 million of cost to the bottom line. Notable about these results is that two thirds of these failures are attributable to system failures which are arguably the most preventable assuming the right solution set is implemented.
These high costs associated with outages coupled with the growing percentage of data centers that are virtualized in large part explains why VMware vSphere offers so many tools to prevent outages and effectively utilize available compute resources. On the availability side, VMware offers VMware HA, vCenter SRM and vSphere vMotion as well as utilities to manage compute resource such as its Distributed Resource Scheduler (DRS) that aligns available resources with the business priorities of the applications running on VMs.
Together these various VMware features help to ensure VMs remain up and operational should an outage occur under most circumstances. What they do not accomplish is ensure that applications remain up and operational under all circumstances nor do they account for the different ways that organizations look to leverage high availability.
For instance, other events that may contribute to a system failure which VMware’s existing tools do not detect or perform include:
- Monitoring the status of the application itself such that if an application fails within a VM there is no way for VMware to re-start the application
- No means to failover from a physical machine to a virtual machine or vice versa
It is addressing these specific gaps in high availability that Symantec’s soon-to-be released version of Veritas Cluster Server addresses. Veritas Cluster Server monitors the application running in a virtual machine and when an error occurs it will fail the application over to a healthy VM, without the requirement of a VM reboot. Additionally, Veritas Cluster Server will continue to protect an application even should an operating system corruption issue occur.
Of these two, probably the most interesting one is the flexibility it affords to move applications from physical machines to virtual machines and back again. Many organizations still cannot cost-justify virtualizing some of their mission critical applications in light of the amount of physical resources that they require. However the physical machines on which these applications reside still need maintenance and updates from time-to-time that may require they go offline in order to be done.
Using Veritas Cluster Server, they get this flexibility to fail over a physical server to a VM during these periods of routine maintenance, such as when a patch or firmware update is applied to the physical server’s underlying hardware. This activity may now take place without incurring an application outage and provides organizations a cost-effective alternative to having to put a second physical server in place to host the application during these maintenance periods.
Another benefit of this approach is that it also gives organizations the opportunity to safely test drive running a mission critical application on a VM. Organizations may be considering moving an application currently running on a physical server to a virtualized environment but are unsure how well it will work.
Using Veritas Cluster Server, they can move the application to the virtualized environment. If it works, great! Conversely, if the virtualized environment cannot handle its work load, they can fail it back to the physical environment without causing an application outage.
No organization really ever wants any of its applications to experience an outage but when it comes to mission critical applications, they are particularly sensitive to expose them to any chance of downtime. This is why many are reluctant to virtualize mission critical applications if they have no means to effectively monitor the application for failures, do a trial run of how the application performs in a virtualized environment, or move it back to a physical environment if required. The soon-to-be released version of Veritas Cluster Server removes the risk of virtualizing business and mission critical applications by protecting organizations from failures that can occur anywhere in the virtualized stack to now include the application layer.