Solution provider takeaway: Solution providers can deliver a great ROI argument around virtualizing customers' disaster recovery site.
In tough economic times, disaster recovery
That said, customers still need to optimize their DR strategy, primarily to free up funding for other projects and to improve their DR readiness. By guiding them toward a virtualized DR strategy, you can address both of those needs -- and enhance your value proposition.
While many, if not most, customers have already started a local virtualization project, only a few have even begun to incorporate virtualization into their DR strategy. The first step in getting them there is to explain the four main cost benefits of virtualized DR: savings from reductions in physical servers, power and storage, as well as from reduced hardware and firmware syncing requirements.
Physical server cost reductions
With a traditional DR plan, customers want an almost mirror image of the primary data center re-created in the DR site. This means a 1-to-1 ratio of servers and storage between the primary site and the DR site, a very expensive capital outlay.
Beyond being expensive, a 1-to-1 server ratio also compromises operations. It can be almost impossible to keep patches at the DR site in sync with the primary site. And changes in storage must be reflected at both sites. Additionally, a hardware upgrade to a server at the primary site almost always also leads to an upgrade in hardware at the secondary site. Tracking the changes and keeping the sites consistent becomes a daunting and expensive task.
Because of these problems, many customers won't maintain a 1-to-1 ratio of servers at their DR site. They'll instead compromise recoverability by selecting a smaller number of servers to be available at the DR site. In the event of a disaster, the remaining servers will have to be ordered and recovered from backup, costing time, money and possibly reputational damage.
Virtualizing the DR site, on the other hand, reduces costs without cutting DR site effectiveness. Running a virtualization application like those from VMware or Citrix at the DR site will enable you to leverage the benefits of server consolidation by creating a remote virtualized infrastructure to host virtual machines (VMs) in the event of a failure. Such a strategy brings significant hardware cost reductions at the remote site.
In primary data centers, most virtualized server environments are now hosting 15 to 20 VMs per physical host. It's not uncommon in virtualized DR to increase the density even further. Being able to consolidate 30-plus physical servers into one physical server can represent a tremendous cost savings for your customer.
Advise your customer to be careful not to take that density to an extreme. There should be enough compute capacity at the DR site so that in the event of a real disaster, your customer will be able to actually perform business functions or have the ability to spawn virtual machines to other systems. The more physical servers there are at the remote site, the higher the utilization can be per physical host. Budgeting compute capacity at the DR sites is unlike budgeting compute capacity at the primary site. The DR site may never have any real workloads on it until the point of a DR event, so make sure you help customers calculate what the full spun-up workload will be.
Power cost reductions
With a non-virtualized DR site, proper power management is a challenge. In the 1-to-1 server configuration detailed above, all the secondary servers would need to be powered on, obviously needing electricity. In reality, most customers won't be able to justify the electrical and cooling costs to have all the DR site servers powered on and waiting for a failure.
It's more common to have the storage powered on and receiving replication data, and a few mission-critical servers powered on with connections to their storage already configured.
While power costs of a non-virtualized DR strategy can be minimized, these workarounds risk recoverability, and a failed DR site recovery can be very costly to the business. Powering on servers that have been off for a period of time meets with mixed results; the server might not power on or be configured correctly. And powering up during a DR event can be a manual, time-consuming process that is very susceptible to human error.
In a virtualized DR model, there are fewer physical servers in the first place. Therefore, it's much cheaper to keep the DR servers powered on, ready, pre-connected to storage and known to be working. This reduces operational costs and ensures that when a DR event occurs, the chances for success are high.
As you know, all the data in the primary site needs to be replicated to the DR site. Under a non-virtualized DR strategy, this is expensive: Millions of files need to be replicated. Tracking those files to make sure that the critical ones are actually arriving at the DR site and are being updated is difficult. And storage needs to be purchased in a 1-to-1 manner.
With a virtualized DR site strategy, all the data (the millions of files) that needs to be replicated and available at the disaster recovery site is encapsulated into a small group of virtual sever images. This greatly simplifies the tracking and verification that all the data is in place, cutting operational costs.
Storage reduction solutions like inline compression and data deduplication also make sense in a virtualized DR strategy. While these solutions will work with most any data, virtualized environments are particularly compressible and have a high degree of redundant data. As a result, the amount of storage capacity required on the remote site can be greatly reduced. When the two solutions are combined, storage can be reduced by as much as 95% at the DR site.
Reduced need to match hardware/firmware
The final area for server cost savings with virtualized DR derives from abstraction of the software from the hardware. As mentioned earlier, a traditional 1-to-1 DR site model typically leads to a pairing of hardware purchases so that the firmware of the systems match.
In a virtualized DR site strategy, since the hardware is abstracted from the software, the firmware between the two sites doesn't need to exactly match. This is particularly valuable in a DR use case because the server hardware at the DR site could be purposely different, meaning that your customer could choose to put either a lower-cost or older servers (repurposed from the production environment) at the DR site.
Two key budget items for your customers are to reduce cost yet maintain or improve their disaster recoverability. A virtualized DR strategy allows you to offer both.
About the author
George Crump is president and founder of Storage Switzerland, an IT analyst firm focused on the storage and virtualization segments. With 25 years of experience designing storage solutions for data centers across the United States, he has seen the birth of such technologies as RAID, NAS and SAN. Prior to founding Storage Switzerland, George was chief technology officer at one of the nation's largest storage integrators, where he was in charge of technology testing, integration and product selection. Find Storage Switzerland's disclosure statement here.
This was first published in December 2008