VMware has announced the inclusion of basic replication technology for vSphere 5 systems, and many organizations are looking at how they can utilize this platform. Currently, tools like Site Recovery Manager (SRM) require the use of approved, third-party array-based replication systems to move VM data and configuration from one physical location to another. In vSphere 5, VMware can still utilize array-based replication, but can also leverage specialized appliances and the vSphere Replication Management Server (VRMS) platform to provide limited replication capabilities.
Most Vision Solutions, Inc. customers are already very familiar with this form of technology, as it is possible to set up this type of configuration using the Double-Take for Virtual Infrastructure (DTVI) product that’s been available for years. The limitations of host-to-host replication have always made DTVI a solid, but secondary choice for our clients; and the same technology is utilized in VRMS, with the same limitations.
DTVI and VRMS both provide solid snap-and-send-based replication from on VMware host device to another; leveraging the VM Backup API to create differencing snapshots. These snapshots are then transmitted via secure shell to the target VM host, where they are applied to the target copy of the VM itself. Command and control systems manage the replication operations and bring the target copy online in the event of a loss of the primary.
This methodology, while absolutely functional, does have some restrictions present in both our own product and the vSphere technology:
- If snapshots are routinely used for other purposes or by other backup systems, conflicts can occur.
- At higher VM densities per host, the overhead of the snapshot systems begins to have a measurable impact on the performance of the guests running on that host.
- Much more bandwidth than necessary is consumed by the requirements to keep latency low and move larger amounts of data for things like Windows paging files and Linux swap-space.
- None of these systems is an useable choice for non-virtualized servers that require DR protection.
In addition, the scalability of ESX’s inbuilt replication, as VMware’s messaging suggests, is aimed at a small environments. For Small- to Mid-Sized Businesses, this is a reasonable limitation, but for larger shops that need more flexibility in architecture, this limit may mean re-engineering your vCluster designs.
The solution is to combine host-based replication like DTVI – where appropriate – with guest-based replication platforms like the Double-Take Availability for Windows and Double-Take Availability for Linux product lines. While they are licensed either by guest or at the VM host level, these products perform true byte-level replication that can easily exclude non-critical files and binaries.
Double-Take Availability also covers non-virtualized production resources, and even has the ability to leverage virtualization for DR where production is physical. One to one configurations from P2P or P2V can be mixed with many-to-one configurations leveraging P2V and V2V tools like the Virtual Recovery Assistant platform included in the software suite.
Since Double-Take Availability is designed from the ground up to scale effectively on any platform, there is neither limit to the number of vCluster nodes, nor the requirement to vCluster at all if individual hosts are a preferred configuration. This universal view also applies to platforms that leverage virtualization tools from other vendors, such as Citrix XenServer and Microsoft Hyper-V technologies.
We’re all for using the best tool for the job, and in this case, we can provide you with all of those tools.
