Application Availability in a Virtual Environment

By Thomas Schmidt

Server virtualization is increasingly common in today’s data center. By implementing server virtualization, businesses realize they can consolidate the application workload of multiple servers onto a smaller number of physical hosts, resulting in improved hardware utilization, fewer physical servers and considerable cost savings.

But virtualization itself creates challenges. Before the advent of server virtualization, a physical server failure would most likely have resulted in a single application outage. In a consolidated virtual server environment, by contrast, a physical server outage has the potential to affect multiple virtual servers and bring down many business-critical applications.

To mitigate this risk, virtualization vendors have built availability tools for use on their platform exclusively. Generally speaking, these can’t provide the level of protection that enterprises have come to expect for physical servers. In particular, these tools do not monitor the health of the most vital component -- the application itself.

Virtualization and disaster recovery
According to a recent survey, 55% of all companies have had to reevaluate their disaster recovery (DR) plans, as a result of implementing server virtualization. For North American companies, the figure was even higher -- 64%. While this is partly the result of server virtualization being used for faster provisioning for DR, it’s also because putting production applications in virtual servers changes the way companies do DR. New tools and processes are needed, and it’s likely that workers will need to be retrained.

But moving from a physical server architecture to a consolidated virtual server architecture shouldn’t compromise application availability. An application’s availability is at risk if any of the components that the application depends upon are not available. For example, network failures can affect an application’s availability. Similarly, a component failing inside an application can affect an application’s availability. Or a virtual server itself could crash, causing application downtime. Whatever the cause, the IT department needs a way to monitor the health of the application and its dependencies, and intelligently move an application to a healthy server in the event of an outage.

However, the introduction of a new server virtualization platform often means the introduction of a new high-availability and disaster recovery infrastructure for protecting applications in virtual servers. According to the survey, 35% of all respondents cited “too many different tools” as the biggest challenge in protecting mission-critical data and applications within physical and virtual environments. Maintaining different tools for physical and virtual environments creates a lot of new complications, including higher training costs, operator inefficiencies, greater software costs and workforces that operate in silos.

In addition, today’s complex applications often span both physical and virtual environments. For example, databases are often kept off virtual platforms because of performance and availability issues. As a result, a pure virtual platform-based approach to application availability may protect a component of an application, but not the entire service the application provides.

Enterprise-class availability
Given the demands placed on today’s data centers, it’s essential that IT managers maximize resources by moving beyond reactive recovery to proactive management of application availability. Proactive management means recognizing risk before it endangers applications and ensuring that applications will failover when needed. Such a solution must be capable of the following:

• Automated failover Can the solution detect faults in an application and all its dependent components, including the associated database, operating system, network and storage resources? When a failure is detected, the solution must be capable of gracefully shutting down the application, restarting it on an available server, connecting it to the appropriate storage device and resuming normal operations. 

• Extensive hardware and platform support Can the solution support all leading operating systems, including UNIX, Windows, Linux and virtual platforms, as well as a wide range of heterogeneous hardware configurations? Using the same solution across platforms reduces training and administrative costs. 

• Automated disaster recovery testing Data center servers and applications are constantly changing, so the regular testing of a disaster recovery strategy is critical to guarantee a successful recovery in the event of a system or site-wide outage. Does the solution enable automated testing that doesn’t disturb production applications? 

• Multi-cluster management and reporting Can administrators monitor, manage and report on multiple clusters on different platforms from a single console? To make management of the high-availability infrastructure easier, companies should look at solutions that provide a simple GUI, visibility across their data centers and powerful reporting tools to provide data to guide improvement plans. 

• Advanced virtualization support The solution must enable the organization to use the advanced features of virtual server technologies, including live migration and workload-balancing tools.

Conclusion
As enterprises implement server virtualization in their data centers and move from a one-application/one-server architecture to a consolidated virtual server environment, application availability becomes more important than ever. For today’s business-critical applications, enterprises require robust, proven solutions. They must be confident that they are protected, not only against physical server failures, but also from outages that affect individual virtual servers and the applications running inside them.

Thomas Schmidt writes frequently about information security topics. He has more than 15 years of experience as a writer and editor in high-tech publishing.