Replication: A DR Solution Made for Virtualization
Replication is a good example of functionality that virtualization simplifies and makes much less expensive.
- By Doug Hazelman
I've often referred to virtualization as a gateway technology, because it's a gateway that allows us to meet our overall business continuity goal with other technologies. Never before in my professional experience have I seen a new technology meet organizations' business goals so specifically and easily as has virtualization.
Virtualization has changed how organizations do business and how businesses are protected, and, while we've done a great job of embracing a number of technologies central to virtualization, our private cloud components (or call them datacenters) still need key operational elements such as data protection and business continuity. Virtualization alone is not enough, but it does have the advantage of bringing functionality that was previously too complicated and expensive for any but large enterprises to smaller organizations.
Replication is a good example of functionality that virtualization simplifies and makes much less expensive. Indeed, replication could very well be disaster recovery's missing link for many organizations.
Historically, replication was a term for expensive storage solutions that required incredible bandwidth and created contentious internal debates about whether to implement synchronously or asynchronously. These were complicated discussions describing what workloads would be subject to this availability technique and who would pay for it. Today, thanks to virtualization, replication delivers more features and is more affordable, making it a disaster recovery option for organizations who have not previously considered it.
Virtual machines are perfect replication candidates because they are much more portable than traditional systems or expansive data profiles on disks. Replication is also another technology that needs to be understood and managed. The end result, however, can add protection levels never before possible at a mere fraction of the previous capital investment.
A replicated virtual machine is a point-in-time copy of a production virtual machine. It is separated by one or more domains of failure such as the host, storage system or location. A replicated virtual machine is materially different than a virtual machine backup because it can be run directly from the host and underlying storage system.
Replicated virtual machines can play a large role accommodating both same-site and whole-site failures. For example, let's say a drive array or entire storage area network (SAN) controller failed and a widespread inventory of virtual machines went offline. Because it's not an entire site failure and other SANs and workloads weren't affected, the replicated virtual machines could be replicated to those unaffected storage systems on the same site for incredibly quick failover, thus alleviating most network and bandwidth challenges. In my professional experience, I've seen IT pros leverage everything from direct attached storage (DAS) to inexpensive network attached storage (NAS) resources for low-cost, same-site replication options that can accommodate that failure scenario.
The story doesn't stop here. Replicated virtual machines can be sent off-site to accommodate entire site failures. This technology brings recovery to organizations without the capital resources required to invest in storage systems, bandwidth and automated-failover mechanisms.
As a general statement, replicating virtual machines brings replication-based failover (same-site or other-site) to organizations that would otherwise not have this option, because the process of getting started has become so easy. In addition, the bandwidth required to do the first full pass of a virtual machine replication job isn't required for an off-site replica. Technologies that leverage seeding, mapping, transfer compression and other efficiency mechanisms make it much easier to start the process.
While replication is an amazing addition to virtualization DR strategies, it is, of course, not the only component. Backup technologies are still needed. With virtualization, backups become more important because stakeholders expect that problems can be fixed just as quickly as new virtual machines can be deployed. Clearly, backup technologies are the right recovery technique for some solutions, while replication technologies are better suited for others.
We live in a world of increased expectations for our private clouds. Data protection strategy also needs higher scrutiny after virtualization. As IT administrators and decision makers, after assessing all possible domains of failure, we know it's possible to line up tools that will create additional protection layers, yet without excessive complexity or cost.
- What about you? Are you wrestling with any of the following questions?
- Whether to leverage replicated virtual machines?
- What criteria make a virtual machine a replication candidate?
- How frequently a replication technology should run?
- What investment is required to make that happen?
These are all critical data protection strategy decisions.
Are replicated virtual machines the missing link in your disaster recovery strategy? Have you started replicating? Share your comments here.
Doug Hazelman is the product strategist for Veeam Software. He frequently presents on behalf of Veeam at industry conferences, and shares his expertise via his blog, "VeeamMeUp" and other social media outlets. Doug advises Veeam customers and partners on best practices and key considerations as they implement and better manage their virtual server infrastructures.