Array-based replication may seem ideal for IT professionals looking to improve their disaster recovery (DR) strategy. It sends an off-host copy of data to a remote site without installing special software in the hypervisor. However, array-based replication is expensive and needs additional components at the DR site for the organization to recover from the disaster fully.
Array-based Replication is Expensive
Array-based replication requires the source and target storage systems to be from the same vendor. In some cases, you can use a lower-end hardware target. Still, you must worry if, when a disaster occurs, the target array will provide the necessary performance to run production workloads. This requirement means that if you bought an all-flash array from a flash-only company like Pure Storage, you must buy an all-flash array for the disaster recovery site.
Look for a solution that abstracts from the hardware and can run on almost any hardware at the disaster recovery site. Be sure to remember that the DR site might become production, so there is a balance between saving money and having something usable. Also, look for a solution that can support flash and hard disk drives and store data on the hard disk drives to reduce costs—yes, hard disk drives are still significantly less expensive than flash. Then when a disaster occurs or during a DR test, promote the most critical virtual machines (VM) and active data sets to flash storage.
Array-based Replication Only Protects Data
In a site-wide disaster, array-based replication only replicates data. It does not replicate the entire environment. Many VMware environments boot the hypervisor from local boot drives in the nodes, not directly from the storage supporting the installation. Also, array-based replication does not replicate the network configuration information to the DR site. IT must account for these independent parts and ensure they have a different data protection and recovery process.
Since Array-based replication only protects data and the target system can only host storage, IT must ensure it has a second, fully licensed copy of VMware active. Dedicated storage systems can’t host VMs, so DR planners must ensure they have servers at the DR site to host the replicated VMs. Because most environments use proprietary switch technology, they also need those expensive switches in both locations. As supply chain issues continue to plague the industry, the old architecture of having just enough servers and network switches and rapidly ordering replacements is no longer viable.
Look for a solution that consolidates the hypervisor, networking, and storage into a single layer. The advantage of converged solutions is especially prevalent in disaster recovery scenarios. However, hyperconverged infrastructure (HCI) solutions still bring complexity since they still have the three distinct layers running as software on each node. One of the advantages of Ultraconverged Infrastructure (UCI) is that the formally three separate layers are integrated into a single layer, greatly simplifying operations, especially setting up a disaster recovery site. This page compares HCI and UCI.
Array-based Replication Complicates Recovery
Array-based replication makes initiating a test of the disaster recovery plan or an actual disaster declaration significantly more complex. The three disparate parts, storage, servers, and networking, must be individually connected. Most organizations don’t have the time or staffing to ensure these connections are made in advance or continually monitored, especially at the remote DR site.
As a result, IT must put the final “glue” in place live during the test or actual disaster, which delays recovery times by days and sometimes weeks. Complexity means that IT will be less likely to want to test the DR process because they don’t have the time to “glue” things together each time they want to perform a test. Again, a solution consolidating these efforts, especially one that integrates networking, makes executing a recovery procedure significantly more straightforward.
UCI is an environment-based replication solution. For example, VergeIO’s UCI solution moves beyond HCI so you can replicate everything to the DR site; all the data, all the server configurations, and all the network settings. There is “one switch to throw” instead of gluing disparate parts together during a disaster.
VergeOS is an alternative to VMware, but if you are still getting ready to switch to VergeOS, its IOprotect capability enables you to continue to run VMware in production while backing up to VergeOS at the DR site.
Since IOprotect runs within the highly efficient data center operating system, VergeOS, it requires less physical hardware at the DR site. The solution also supports flash and hard disk drives, so IT planners can leverage the 10X cost advantage but promote workloads to flash when disaster strikes. The efficiency, the ability to use existing hardware and the support of both flash and hard disk drives mean that IT planners can lower DR costs by over 50% while simplifying the DR process. The cost savings are so compelling customers are switching now even if they have a couple of years left on their licensing.
The affordability of the IOprotect solution enables many customers to use it to protect against minor disasters like ransomware or storage system failure by deploying an on-premises DR system and then replicating it to their remote DR system. Again the on-premises system can run on existing hardware and leverage hard disk and flash drives.
IT needs to simplify DR and make it more affordable. HCI is a good start because it eliminates some of the physical components that need to be purchased at the DR site, but it still requires proprietary network hardware or VMware’s NSX, which can be just as expensive. It is also complex to operate because IT still works with three separate tiers. UCI is a superior solution that consolidates everything, protects VMware, and dramatically lowers the cost of DR.
To learn more about creating a VMware DR solution that lowers DR costs by 50% and simplifies recovery, watch our virtual whiteboard panel discussion, “Whiteboard Wednesday: VMware Disaster and Ransomware Recovery—The Three NEW Best Practices”.