This content is part of the Essential Guide: Take a chance on virtualized backup and disaster recovery

Build a remote data replication strategy

Synchronous and asynchronous remote replication provide different options for IT. Understand the differences, and the potential effects on your network, to keep enterprise data safe.

For workloads, protection means resilience -- ensuring that an application is available. But even more important is the protection of enterprise data. IT can always reinstall or spin up new iterations of applications. However, there is no ready substitute for data that becomes inaccessible or lost.

A remote data replication strategy can help IT professionals protect vital enterprise data, and there are numerous alternatives to consider.

Q. How does remote data replication work?

To perform remote replication, IT teams capture a copy of production data, and then save that copy to a location other than the primary data center.

IT teams can replicate data from a primary data center to one or more secondary data centers, an outsourced service provider or to the public cloud. The final destination will depend on cost, as well as availability and required control. Also, consider what will drive the replication process; replication can occur at the host level, from within the storage array and from within the network.

To create a data replication strategy, IT leaders must also decide whether they copy data synchronously or asynchronously. Synchronous data replication ensures that both primary and secondary data writes are all completed successfully, eliminating the potential for data loss between primary and secondary sites. Synchronous replication offers the smallest recovery point objective (RPO) and recovery time objective (RTO), but network bandwidth demands can be significant.

If remote replication demands even greater distances, or insufficient bandwidth is available to support it, consider using asynchronous replication. Asynchronous replication essentially uncouples the primary and secondary writes, allowing writes to the secondary site(s) to be completed later.

Q. How does the network affect a remote data replication strategy?

Where local replication depends on a local area network, remote replication relies on a wide area network (WAN). The WAN can be a dedicated network, but is typically a shared IP network, such as the internet. This means any issues that affect network availability and performance will also affect remote replication and recovery.

Network outages can prevent replication or recovery processes. Dedicated networks, or the use of multiple telco providers, can provide superior reliability.

There must also be adequate bandwidth to convey the required data copy from the primary data center to the secondary location within a reasonable timeframe. For example, moving 10 GB of replicated data over a 1 Gbps connection ideally takes less than two minutes, but replication rarely ever has exclusive control of the network. Each business application still needs to function with little performance penalty during the remote replication process, so the actual time to copy data may be much longer.

Bandwidth isn't so critical for asynchronous replication, but it can seriously affect synchronous remote replication performance. Aggregate dedicated or shared WAN links to provide adequate bandwidth. Admins can also use mitigation technologies, such as data deduplication, to reduce bandwidth requirements.

Data security considerations

While admins may not need to encrypt local data storage, they may need to encrypt the data copied for remote replication before it goes out onto the WAN. That data should remain encrypted at rest while stored at a remote location. Effective encryption may be a critical part of corporate governance and regulatory compliance.

Q. What are some characteristics of remote data replication tools?

There are a number of tools that can help support a remote data replication strategy, but IT pros should consider their requirements before choosing one.

First, weigh the impact of the remote replication platform on your production environment and consider a replication tool that minimizes any additional I/O burden. For example, a host-based replication tool such as Quest vRanger or Double-Take DR by Vision Solutions will need to run on a server, and this will demand compute power and disk I/O to complete replication tasks. By comparison, a remote replication appliance can typically copy writes in-line within the SAN, which essentially eliminates any performance impact on servers and storage.

Where synchronous remote replication always mirrors writes from a storage source to a storage target, asynchronous remote replication can experience a delay between source and target writes. This means asynchronous replication demands a technology that can accommodate RPOs and RTOs. Continuous replication technologies such as Zerto's hypervisor-based replication tool or Xen Orchestra are two potential options.

Look for journaling capabilities that allow administrators to recover specific points in time. This kind of feature can provide more flexibility and granularity than simply retaining and recovering the latest data replica, but demands some additional storage and I/O.

Shoot for integration between the remote replication platform and enterprise disaster recovery or systems management tools. For example, operating the remote replication platform through tools like VMware vCenter Site Recovery Manager or Microsoft System Center reduces the number of tools in the data center and simplifies management.

Q. How does a remote data replication strategy compare to other backup options?

Remote replication remains an important data protection technology, but there are some alternatives organizations should consider.

First, there are more options available for target storage -- especially in the cloud. Enterprises can copy data to storage instances in public clouds such as Amazon Web Services, Microsoft Azure and Google Cloud Platform. Users only pay for what they need, but there's a fee when data moves into and out of cloud provider platforms, which can make extensive recovery operations costly.

Another option is a local replication target, such as the hardened Black Box from Axxana, which is built to withstand physical challenges, such as a fire. However, such on-site replication options don't suit all disaster scenarios, and organizations may need to use them in conjunction with asynchronous remote replication options.

Lastly, admins can use virtual machine snapshots to capture the state of a VM running in server memory at any point in time. Admins can usually accomplish snapshots in seconds without quiescing most applications, and they have relatively little I/O burden. Admins can then replicate the snapshot remotely as needed.

Next Steps

Data replication software tools boost redundancy

Replication fuels disaster recovery purchases

Redundancy plays a major role in data management

Dig Deeper on Enterprise data storage strategies