Amid all the talk about virtualization laying the groundwork for cloud computing, dynamic resource allocation and other grand schemes, most enterprises are barely scratching the surface when it comes to such mundane activities as server consolidation.
As with any new technology, the promises are great, but so are the pitfalls if you rush into it too fast. So it seems likely that many organizations will pursue the big picture through a series of baby steps -- testing virtualization's promise in a range of more finite applications before undertaking wide-scale infrastructure reorganization.
One of those applications seems to be disaster recovery. Always a major cost center with a very low return (until you need it, that is), DR stands to benefit greatly both in cost of deployment and overall performance through virtualization.
One of the chief advantages to virtualized backup is the ease of creating off-site environments that can be switched on in time of need. That's one of the prime features in the StorageLink Site Recovery module of the new Citrix Essentials package for Hyper-V. The system provides a single console for creating and testing secondary instances, essentially providing a common platform that traverses both the primary and back-up site so workloads can be easily transferred should the primary go down. The entire package can also be managed through Microsoft Systems Center if needed.
EMC users running VMware also have a new-found ability to set up off-host backups for virtual machines. Now that EMC has integrated its Retrospect B&R module into VMware's Consolidated Backup system, the package can be set to automatically alert idle workstations to accept back-up data, regardless of whether they are running Windows, Vista, XP, Server 2008 or Server 2003. A new Retrospect Emergency Recovery CD can then be used restore downed images.
Virtual DR is valuable, but it certainly is not a simple plug-and-play proposition. According to Don Norbeck, director of product development at SunGuard Availability Services, the complexity goes up astronomically depending on how mixed the primary environment is. A single DR platform is crucial, but it must be capable of restoring all the physical and virtual environments that users rely on to prevent simple outages from seriously hampering productivity -- and the bottom line. To do this, it will most likely be necessary to architect a solution that maps multiple resources down to fewer hosts.
Without that "many-to-one" approach, you'll likely have to keep an eye on your storage capacity, according to Network Computing's David Greenfield. Systems that create virtual machines as part of the back-up process (he uses Acronis as an example) can aggravate the problem of virtual sprawl that is already starting to plague virtual environments. Adding data-reduction techniques like deduplication and compression can help, but be aware that each bit of data that finds its way into back-up, whether virtual or not, will be replicated multiple times across the storage tier.
But even if storage requirements do go up, virtual DR is a worthwhile investment simply due to its ability to get users up and running within minutes, rather than hours or even days. After all, the most expensive proposition of all is not being able to function.