Replication is an essential part of the data protection strategy for many organizations. However, before an organization can benefit from data replication, it must determine whether it's better to perform hardware replication or to replicate data at the software level. There are advantages and disadvantages to each approach.
Many enterprise storage vendors integrate hardware-level replication capabilities into their devices. Such devices are linked by a high-speed network connection. When data is written to the primary storage device, the device automatically replicates the operation to a secondary device. This replication occurs as a function of the device firmware and isn't dependent on an external software application.
Some storage vendors design their hardware to perform block-level replication. The primary storage device keeps track of the blocks that are written or modified and then replicates those blocks to the secondary device. However, modern storage hardware is increasingly being designed to replicate snapshots rather than individual storage blocks.
Hardware replication tends to be best suited for situations in which data needs to be replicated either between two data centers or between two identical storage devices within a single data center. Because hardware replication is hardware-dependent, it tends to be a poor choice for those who are on a budget.
Some storage devices do provide the option of replicating the array's contents to a public cloud, but capabilities tend to vary significantly from one device to the next.
Software replication comes in many different forms but is based on the idea of using an operating system, a hypervisor or a mission-critical application to replicate data from one location to another without the replication process having to be directly supported at the hardware level.
Microsoft, for example, has built a replication feature into Hyper-V that enables VMs to be replicated to a secondary system. If the primary system fails, then the replica VM can be activated as a way of bringing the VM back online.
Similarly, many applications perform a type of database replication called logical replication. Logical replication keeps track of which objects have and haven't been replicated, based on the object's primary key. As objects are created or modified, the database replication engine migrates those changes to a replica database on the target system.
Many backup applications also include a replication feature. Backups are initially written to a designated target. That target is then replicated to a secondary location, either on premises or in the cloud, as a way of creating a redundant backup copy.
Software replication tends to be a good choice for those who require application awareness and the flexibility to create multiple replicas in a variety of locations, although not every software-based replication engine supports multiple replicas.
It's also a good choice for organizations that don't want to have to purchase matching storage hardware. If a replica only exists for the sake of redundancy, for instance, it might make more sense to write the replica to commodity storage than to store the replica on a high-end array.
Neither hardware- nor software-based replication is an ideal approach for every organization. Each approach has advantages and disadvantages that must be considered prior to making a purchasing decision. When deciding on a replication approach, organizations must consider factors such as their budget, the number of replicas required, existing hardware, the type of data to be replicated and the location of the replicas.
Of course, choosing a data replication type doesn't necessarily have to be an either/or decision. It's possible to implement a hybrid replication strategy that uses both software and hardware replication. For instance, an organization might use software replication to replicate a VM to a secondary data center, where the VM is then replicated to an additional storage array using hardware replication.
01 Jul 2020