In the recent past, Society Insurance reviewed its business-continuity capabilities in the event a disaster struck. It found it would need a full month to execute all the steps required to restore normal business operations, including acquiring and deploying replacement servers, reinstalling its applications and reloading data while operating at normal staffing levels.

Society Insurance officials realized immediately that the time interval was untenable. Consequently, they set out to update their technology infrastructure for a vastly higher level of business continuity.

The firm's specific goal was a maximum two-day recovery timetable. Importantly, the company conducted this analysis on a proactive basis, not in response to an outage or disruption.

"Upper management was saying it's not acceptable to be down or have a lot of critical applications down for multiple days in a disasterrecovery situation," explains Tom Konop, Vice President, Information Systems for Society Insurance. "That drove us forward to say we need to improve."

In its 30-day disaster-recovery scenario, Society Insurance would have relied on transferring data processing and application hosting to a third-party facility. Recovery activities would have required physically taking servers and software onsite and setting them up at that location.

"That didn't exactly make for easy, efficient testing of disaster recovery processes," Konop says. "And it certainly contributed to the lengthy time lag before normal operations could resume."

The firm's existing infrastructure could have supported basic business functions such as conducting inquiries. However, it would not have allowed employees to readily perform underwriting, claims processing and other core insurance functions in a timely fashion as their jobs, and customer needs, require.

The company considered blade servers as an option to provide additional flexibility and speed in terms of resuming normal operations. Society Insurance also considered replication and recovery solutions based at the operating system level on each server, for example NSI Double-Take.

In either case, the technology didn't provide the desired reduction in physical equipment and infrastructure. "It would have been close to a one-to-one replication" of hardware, Konop explains. Notably, more physical equipment also means more costs at the disaster-recovery location. And that means undesired additional overhead and complexity for business continuity.

Instead, the company's IT leaders opted to deploy server virtualization. The technology allows the firm to streamline and enhance infrastructure while at the same time increasing manageability and speed when it comes to Business Continuty (BC).

Server virtualization aggregates data center resources, including processing, memory, storage and network to support virtual workloads. It also allows the sharing of physical resources to increase overall utilization, the reduction of costs through hardware consolidation, and provides a hardware-independent platform which provides flexibility in recovery through the use of virtual machines.

From an administrative standpoint, this means easier management. IT professionals can manage these virtual machines, both the system state and data, as objects which can be easily moved and replicated.

The technology also lets them respond dynamically to changing needs. These could be temporary needs for more storage or server capacity, or even a need to move applications to another system because the one they were running on has failed.

Virtualization is widely viewed as a technology that can help companies more quickly recover from disaster and resume normal business activities. It's also frequently viewed as a way to reduce IT infrastructure costs.

Society Insurance deployed virtualization in the data center of its home office in Fond du Lac. VMware Infrastructure 3 (VI3) and ESX Server were chosen as the virtual platform. For storage, the company is using Storage Area Network (SAN) technology from NetApp.

The company not only built the virtualization into its primary data center, but also replicated all functionality at CDW's data center in Madison, Wis. – about one hour away – where Society Insurance leases rack space. The remote data center serves as the key link in the firm's disaster-recovery strategy.

So how does virtualization facilitate disaster recovery for Society Insurance? Several ways: A virtual machine, including the operating system(s), application(s) and data, is "encapsulated" in a set of files. These reside on a disk-storage system within a shared SAN.

The files themselves are replicated at the block level as individual blocks change. This has a dramatic reduction in the amount of data being replicated. In addition, NetApp replicates the data via Transmission Control Protocol/Internet Protocol (TCP/IP) natively so additional SAN Fibre Channel over IP (FCIP) hardware is not required for replication.

"You take a server, put it in a file, now the file can be transported anywhere you've got a connection," explains Seidl. "We present that to a host in the disaster-recovery site and start the server back up. That's the real meat of it. Plus, we've got tools that help with the process."

One of the most critical of these tools is "snapshot" software from NetApp. It quiesces the current state of the virtual machine and all associated data at a given point in time.

Society Insurance is "snapping" its servers every four hours using NetApp's SnapManager for Virtual Infrastructure software or SMVI. The software also initiates the replication of the data which again is done at the block level.

"We picked that timeframe because it's a matter of balancing the time to transmit all the data," Konop says. "That means longer time intervals would result in more data and more time to upload that data to backup systems."

With a typical backup approach, you could lose up to 24 hours worth of data after a disaster. The four-hour snapshot window reduces the data loss and therefore the impact to the business.

That's a big relief to the firm's business users across the company. They can now operate with a high level of confidence knowing that any interruption in their ability to sell and service policies will be minimized.

With the exception of servers that can't be virtualized due to unique hardware requirements, fax servers, for example, all other applications at Society Insurance now run on virtualized systems with the exception of Microsoft Exchange. The firm's Exchange environment will be migrated to Exchange 2007 in their virtualized environment when that upgrade project is undertaken later this year.

In the company's virtual infrastructure configuration, it hasn't built a one-to-one replication between the hardware used as server hosts in its production and disaster-recovery sites. Its production site has three physical server hosts while the disaster-recovery site has just one that is used specifically for virtualized operations.

It was determined that the one server host could be used to bring up business-critical systems in a business-continuity situation. For a longer-term situation, additional server hosts would be acquired accordingly to bring up non-critical production, development and test systems.

"This allows us to do all the testing we need while bringing up the environment," Konop explains. "In a time of disaster, there may be some concessions in that all applications don't run in a DR [Disaster Recovery] situation."

For instance, software development and testing servers plus accounts receivable data may not come up or be immediately available. However, they are replicated and can be brought online if circumstances require doing so.

"Virtualization simply gives us more options on the DR host," says Konop. "Because not all of our production applications require 24x7 uptime, we can strategically swap out critical servers as needed to keep the business running."

The firm's business-continuity plan specifically prioritizes applications and the order in which they should be brought up. This is based on business necessity.

Priorities are determined by two primary factors: interdependencies between applications as well as importance to the business.

That prioritization is reviewed regularly and can be updated if and when necessary. "By allowing for less than one-to-one replication, we save money on hardware at the DR site," Seidl adds, "And while our production site grows, we still have incremental capacity available at the DR site." The company also achieves some economies at its remote site by using lower-cost networking and storage gear. For instance, it's using a Fibre-Channel SAN at its primary data center but using a lower-cost iSCSI based SAN at the business-continuity data center, according to Gerry Thome, a solution architect at CDW, who worked with Society Insurance on its business-continuity strategy.

With the implementation of server virtualization, Society Insurance has eliminated many physical boxes in favor of virtual machines. While cost savings wasn't the primary driver for the firm, simplicity is clearly a huge benefit, and one that serves business-continuity objectives as well.

Simply put, server virtualization allows businesses to combine and consolidate workloads on a smaller number of physical servers. The main goal is to reduce costs and increase hardware utilization. Basically, it allows a firm to maximize the value of their IT infrastructure investments. That can help increase ROI as well as cut the Total Cost of Ownership (TCO).

Society now runs roughly 50 virtual servers inside three physical systems. The company previously had at least three dozen physical servers and would have continued to add more for new applications and business requirements.

IT officials at Society also report that virtualization makes it easier and quicker to provision new virtual servers when compared to doing so with physical servers. Virtual servers can be backed up faster and more easily by "snapping" a server for backup purposes. This capability makes server upgrades less risky as well.