Many IT managers are considering moving their production applications to virtual server environments. However, with all the benefits that come with virtualizing, such as increased server utilization and quicker server provisioning, companies also find an increased risk of application downtime.

High Availability (HA) offerings from server virtualization vendors have many limitations as HA is not their core competency. As a result their customers are turning to outside vendors to obtain enterprise-class High Availability and Disaster Recovery (HA/DR) - especially for data center applications.

How can IT managers get the same level of high availability and disaster recovery protection they expect in the physical world in the virtual one? And what are the HA/DR requirements for a mission critical enterprise application?

ENTERPRISE CLASS HIGH AVAILABILITY AND DISASTER RECOVERY

A mission critical application is an application that needs to be running within 8 hours or less of an incident that causes downtime. The first step in building a HA/DR solution is monitoring the status of the mission critical application. Is the application running? If it isn't, the administrator should know immediately. In the physical world, this means monitoring not just the application, but all of its dependencies including the application components, the operating system, the physical server, the network connection, the storage connection, and the health of the entire data center.

Besides monitoring the application and its components, another key aspect of a HA/DR solution is taking the appropriate action once an application has been found to have failed. If any of the components that are being monitored have a fault, the appropriate action is automatically taken, i.e., the individuals responsible for the application are notified of the outage and the recovery of the application begins. In the physical world, IT organizations use high availability clustering software to monitor and restart the application so that end users can continue their normal operations as soon as possible. In case of a regional disaster or other event which may take down an entire data center, companies have put in HA/DR solutions that ensure that data is continuously replicated at a second site, and using clustering software automate the process of starting up applications at the second site and connecting users to these new application instances.

Regardless if the application is running in a physical or virtual environment, service level expectations for enterprise applications do not change. However, there are some new challenges facing IT teams in meeting these business requirements.

CHALLENGES IN A VIRTUAL SERVER ENVIRONMENT

To reap the benefits of server consolidation, companies are taking on more risk without plans to mitigate those risks or an understanding of what they are. One reason is that more applications are now on fewer physical servers. In the physical world, a physical server failure may impact the few applications that are running on that server. In the virtual world, there could be 10 or 20 applications on a single server. The failure of a physical server running server virtualization technology impacts many more applications than it would have in a typical physical configuration..

Another reason for the increased risk is now there is another layer of technology that needs to be managed and monitored for potential outages. IT administrators still have to manage and monitor the system components they would in a physical environment, but in addition, IT professionals now have to manage and monitor the virtual server and the virtual server infrastructure. These management and recovery solutions for virtual infrastructure may be different than it would be in physical environments. Tools used for recovery of physical servers may not be appropriate for virtual servers and may have different functionality and different management interfaces. The result is the server virtualization technologies become an additional platform for IT teams to manage and, in turn, build specific HA/DR solutions for. Building a new HA/DR infrastructure brings increased hardware and software costs, but also leads to a more siloed work staff and greater operator inefficiencies as workers need to learn or create additional tools, log into more management consoles, and get their arms around a more complex IT environment.

VIRTUAL SERVER HIGH AVAILABILITY AND DISASTER RECOVERY OFFERINGS

Leading server virtualization vendors currently offer a variety of products that claim to offer customers solutions that address some basic HA/DR requirements.These server virtualization HA solutions are typically bundled with a lot of other features, and are relatively inexpensive and easy to implement. They typically protect against a physical server failure by restarting the virtual server on another physical server..

However, they do not provide visibility nor monitor any applications, application components, the virtual machine itself, the network connection, the storage connection, or the status of the data center site itself. For some applications, this level of protection may be sufficient, but for mission critical applications, most IT organizations will require a greater level of protection and will seek another solution.

Many customers consider their virtual machine live migration tools, i.e. tools that maintain the state of the virtual server during a move from one server to another, as a key component of their High Availability plan . These solutions do not have application monitoring capabilities, and require both servers to be operating for a virtual machine move to take place. While these type of tools are excellent for planned maintenance, they do not take into account unplanned downtime or issues when one of the two servers may be down. As a result, they provides a great product for prevention of planned downtime, or within a virtual staging environment, but have limitations in terms of a HA solution.

In case of a site failure, companies develop a disaster recovery strategy that will help organizations migrate the IT operations of their production data center to their secondary data center. For many organizations in virtual environments, this is likely a manual process. While data may be replicated to the second site on a continual basis, there is not a process for monitoring the health of the site and definitely not the capability for automating its recovery at the second site. Thus, most DR solutions for virtual environments require a great deal of manual work and expertise that may not be available during an actual disaster.

ENTERPRISE CLASS HIGH AVAILABILITY
AND DISASTER RECOVERY FOR VIRTUAL SERVER ENVIRONMENTS

When choosing a High Availability and Disaster Recovery solution for virtual server environments, IT organizations should ensure that the technology they implement meets certain criteria. Most importantly, if a company is placing mission critical applications in virtual servers, they must ensure that they have an enterprise-class HA/DR solution that:

monitors the application and application resources, including the virtual machine, network components, storage components, and the physical server,
notifies administrators of failures to any of those resources, and
automates the recovery and start-up of applications including reconnecting users to the restarted application.

Without these critical solution components, your mission critical applications are not being adequately protected.

Second, organizations should consider if they want one or multiple solutions for both their local high availability and remote disaster recovery plans. Does the organization have a separate plan, processes, and infrastructure for disaster recovery? An appropriate HA/DR solution for virtual server environments should work with the existing disaster recovery infrastructure, i.e., server and storage platforms and replication technologies. Testing of an organization's DR plans is critical to ensuring that it will work when needed. A DR plan should also allow for easy testing of the DR plan without impacting the production environment. Testing should be automated, isolated, and measurable.

If the goal is to simplify an IT environment, then IT organizations should look at HA/DR solutions that excel in supporting both physical and virtual server environments thus increasing administrator efficiencies. In this case, an ideal HA/DR solution should provide the same functionality and a single management interface for their entire HA/DR infrastructure regardless of the operating system, virtualization technology, underlying server and storage hardware, or location of the application. Often companies build point solutions that address each specific application or platform. However, by standardizing on a single platform for both physical and virtual environments, organizations can reduce training costs and increases worker flexibility, as well as reduce administer burden by providing a single console for making configuration changes, moving applications, and reporting.

Finally, when choosing a HA/DR solution, IT organizations should seek solutions that are compatible with the advanced virtualization features of the server virtualization vendors that provide organizations with tools for planned maintenance and work load balancing. IT managers should only purchase HA/DR technology if it allows an organization to take advantage of all of the functionality of the server virtualization technology. An HA/DR solution should complement and enhance an IT environment by providing greater control and visibility, as well as increased availability.