As server virtualization takes over the datacenter, the glitz and glam of the rich features within the VMWare suite easily overshadows the stodgy world of server hardware that it operates on. This is understandable, however, the perception that server virtualization diminishes the importance of the hardware layer is entirely out of focus. On the contrary, virtualization of servers increases the importance of the hardware layer for the simple reason that the ability to host multiple virtual machines amplifies any hiccup in hardware in proportion to the increased number of services the hardware now supports. In the one-to-one world, where a single host provided a single service, the impact of hardware failure could be gauged in a linear manner. With hardware today supporting a many-to-one relationship with virtual machines, a failure could result in the loss of multiple services, and damage to many different datasets. For this reason, VMware developed a feature known as High Availability (HA) and included it as an option in the vSphere product line.
HA allows a cloud service provider to deliver value from the cloud by provisioning extra capacity within their hardware cluster that can be used to maintain services in the event of a failure. For instance, the provider might provision a cluster of three physical servers, or hosts, to support a group of 9 virtual machines. With HA enabled, VMWare reserves a portion of each host to act as failover for the other units. Should hardware become unavailable, say to a power or system board failure, HA will recognize the loss of the host, and re-allocate the three virtual machines that the unit supported to the space on the two remaining functional hostst that it has reserved for this purpose. This decoupling of service from hardware reduces the impact of failures, and maximizes server and application uptime.
Anyone having had the pleasure of configuring a clustered environment at the application layer will attest to the difficulty this task has presented in the past. The beauty of HA is that the complexity of providing an environment with built-in application failover is handled at the server level by the provider. No more dual operating systems requiring twice as many patches, and twice as many potential problems. With HA, an autonomous server hosts an application, and in the event of hardware malfunction, that same server is maintained, and continues with the dedicated services that it’s configured to provide.
The net result of obtaining a service from a provider that has implemented HA in their virtual environment is that high availability for applications is achieved in a manner that is easy for the client. All an organization need know about HA to ensure that their applications and services will run with the highest degree of availability is that it is implemented on their provider’s platform. With this knowledge in hand, those responsible for service delivery are assured that the best possible uptime is delivered.