Business Continuity/Disaster Recovery
Industry experts say virtualization increasingly will be used to support business continuity and disaster recovery. Clearly, with so many business operations tied to servers, keeping systems and applications available to both internal users and external customers is a high priority for IT executives.
IDC estimates that server downtime cost organizations about $140 billion worldwide in lost productivity and revenue in 2007. Because virtualization software effectively "decouples" application stacks from the underlying hardware, IDC says, a virtual server can be copied, backed up and moved just like a file.
The firm says a growing number of virtualization software vendors have incorporated the ability to support live migrations. This, plus the decoupling capability, provides a low-cost means of quickly reallocating computing resources without any downtime.
The key benefit of virtualization in business continuity/disaster recovery is the ability to deliver on service-level agreements and high quality of service, IDC's Gillen says. The main challenge is that organizations need some level of redundant infrastructure and the ability to leverage that infrastructure.
"In other words, you can't virtualize one server and get business continuity," he explains. "You need to have multiple servers provisioned and equipped to address this use case."
Marriott International has launched a virtualization effort with the goal of virtualizing 100 percent of its servers. One of the main drivers of this strategy is to improve flexibility and support the company's business continuity and disaster recovery programs, says Dan Blanchard, vice president of enterprise operations.
Marriott has about 45 percent of its Windows and Linux servers running in a virtualized environment, Blanchard says. Virtualization software from VMware (for Windows servers) and Red Hat (for Linux) lets Marriott recover applications and data faster and more reliably, he says.
Business continuity and disaster recovery--as well as capacity planning--motivate his team to use virtualization. "Our virtualization strategy is certainly to save money," Blanchard says. "We're increasing our utilization of servers. But a big part of why we're doing this and have a 100 percent virtualization objective is for that flexibility. We [intend to] have a virtual layer between [all] our applications and servers"--even for servers that are used for a single application.
Business continuity and disaster recovery are functions that "traditional workload management solutions don't address quickly, efficiently or reliably," Blanchard says. "By doing virtualization, we are able to move workload in reaction to any [events] and recover services or move them more quickly than we've ever been able to do in the past."
Marriott operates several server farms within its data centers, and the servers in those farms are being virtualized. The company is constructing a new underground Recovery and Development Center (RDC) (which was scheduled to open in January) to handle workloads in the event of a service interruption in its main data center.
"When it becomes a business recovery problem, we are positioned to move workloads between the centers," Blanchard says. "One of the best advantages virtualization provides is that the disaster recovery process is the same process we use day in and day out to manage the workload within the data center, so we have much higher confidence in case of a disaster."
Marriott is building the RDC not only to provide a disaster recovery site, but also to accommodate future business growth. The company uses that location to do development work, Blanchard says. With virtualization, it can move development workload off the RDC systems and move production work onto those systems as needed.
Huntsville Hospital uses server and desktop virtualization for business continuity. The organization's IT environment consists of IBM blade centers attached to an EMC SAN. The hospital uses this environment for both virtual desktops and virtual servers.
VMware software has features called Distributed Resource Scheduler (DRS) and High Availability (HA). DRS monitors host systems, and if the resources of a host are overextended, DRS uses another feature called VMotion to automatically move less utilized virtual "guests" to another host.
HA serves a similar function, only it looks for a host that is not running. "If a [host] has a hardware failure and is down, HA will bring down the virtual guests on that host, move them to another host or hosts, and restart them," Wilburn explains. "Since all this happens automatically, the amount of downtime involved is negligible."
Host systems are spread over two data centers. "Because of this, as long as we don't lose the SAN, we could lose a whole rack of servers and still stay operational," Wilburn says.
Another key feature of the VMware software is Site Recovery Manager (SRM), a tool to manage failover in a disaster recovery situation. With SRM, which Huntsville is evaluating, organizations can manage which virtual machines fail over to which site, as well as the order in which they are brought online.
"This keeps you from having to remember the order in which systems have to come online when you're in the midst of a disaster, as well as letting you test the failover without an actual failover to find the best scenario for your organization," Wilburn says.