While enterprises have traditionally used the wide area network (WAN) to link their remote sites to corporate headquarters and to the Internet, the WAN can also play a significant role in an organization’s disaster recovery and business continuity planning. A wide area network disaster recovery plan, built upon the higher speeds and better availability offered by today's advanced WAN technologies, gives enterprises more options for replicating data and establishing disaster recovery sites.
Off site replication becomes a reality
MPLS networks offer WAN managers a number of compelling features to support a new wide area network disaster recovery plan. First and foremost, MPLS and Metro Ethernet offerings deliver significantly higher speed connections than more traditional WAN offerings, making site-to-site replication a reality. Pushing data from the primary data center to a secondary location within the enterprise could be an attractive disaster recovery (DR) alternative to moving backups off-site with tape or disk media.
MPLS is also a mesh technology, which eliminates the single point of failure associated with point-to-point WAN technologies. Most traditional WAN networks, such as ATM or T1 links, rely on a hub and spoke topology with remote sites connected to a centralized location. If a disaster knocks out the central hub, all of the remote sites lose connectivity to the central data center and to each other. An MPLS network gives enterprises the resiliency of a true mesh network, with any location capable of connecting to any other site on the network. With no central point of failure, a disruption to connectivity at one site will not affect the connectivity of the rest of an enterprise's sites. Also, by establishing a failover or alternate data center on the MPLS network, an enterprise can quickly redirect users if disaster befalls the primary data center.
What to know before you start a wide area network disaster recovery plan
Before sitting down with data center managers, application managers and storage administrators to talk about changes to their organization’s disaster recovery program, WAN managers need to know exactly what their wide area network is capable of.
“The primary takeaway for WAN managers regarding disaster recovery is to always make sure the WAN link is working, no matter where that link needs to be,” said Deni Connor, principal analyst at Storage Strategies NOW.
If an enterprise is exploring where and how it could establish a disaster recovery data center, WAN managers will need an inventory of the capacity, performance and interfaces used for each link on the WAN. Likewise, if an enterprise plans to use a third-party provider for hot site relocation, the WAN manager will have to work with both its carrier service providers as well as the DR vendor to ensure that the hot site can be quickly linked to the rest of the enterprise’s wide area network during a disaster declaration.
A mix of replication methods and WAN optimization
In a wide area network disaster recovery plan, the WAN must not only be able to deliver a fast failover to a secondary data center, but it must also enable replication of data to that disaster recovery site. In addition to being faster, new WAN technologies are more reliable, enabling enterprises to use a mix of asynchronous, semi-synchronous and synchronous replication when moving data between the primary data center and secondary recovery sites.
Synchronization methods differ in how frequently the receiving system sends an acknowledgment to the sending system that it has received the data. In synchronous replication, the secondary site acknowledges nearly every change. Because of the send/acknowledge cycle, this is regarded as both the safest and slowest method of replication. Semi-synchronous replication still sends acknowledgement, but only after a series of changes have been sent. And finally, asynchronous replication relies on a highly reliable connection, as the primary system simply sends the replication changes without expecting an acknowledgement. Asynchronous replication is the fastest method of replicating a system, but also poses a higher level of risk because there is no assurance that the system at the other end of the line ever received the change. Today's advanced WAN technologies mitigate this risk.
“Higher quality WAN links are allowing organizations to take advantage of the performance that comes with asynchronous replication by minimizing the risks associated with the technique," said Connor of Storage Strategies NOW. “Enterprises should identify the criticality of the application, the required recovery time objective (RTO) and recovery point objective (RPO), and select a synchronization method that balances all of these with the associated costs.”
Disaster recovery replication can burden even the most robust, modern WAN transport technologies. WAN optimization technology can reduce the impact of the backup and replication traffic between sites. These products use a combination of compression, caching, and protocol optimization to improve the performance of transfers across the wide area network. Likewise, in some cases, WAN optimization appliances can clear up hidden packet loss problems, improving the network performance for both backup and primary application usage.
WAN optimization products usually require an appliance on both ends of the WAN link. While this is easy enough to accomplish when optimizing the connection between two enterprise locations, getting a matched set of accelerator appliances between the enterprise and a third-party disaster recovery site or cloud provider could pose some challenges. WAN managers should make sure that WAN optimization options are discussed early on during negotiations with third-party disaster recovery providers to ensure a compatible solution is available.
Cloud-based options for a wide area network disaster recovery plan
Cloud computing disaster recovery solutions will emerge in the next few years, giving enterprises more affordable and flexible options for storage backup. These cloud-based disaster recovery solutions will require the support of a robust wide area network.
“The wide area network is a pathway to the cloud," said Jeff Boles, senior analyst for the Taneja Group. "Through virtualization and cloud solutions, disaster recovery op tions that were previously cost prohibitive are much more accessible to small and medium enterprises. With the bandwidth and connectivity options available for enterprise WANs, moving these large volumes of data between sites or into a cloud solution becomes possible.”
“The bandwidth situation has changed dramatically, said David Sobel, CEO of Evolve Technologies, a Virginia-based managed service provider. "While tape is cheap and still practical for storing massive amounts of data, cloud storage is becoming as cost effective for enterprises of any size.”
Cloud services could also give enterprises a viable alternative to maintaining a disaster recovery data center. An enterprise could replicate its virtual environment to a multi-tenant cloud provider at a cost much lower than traditional hot site service providers.
Evolve Technologies already offers this capability to its cloud storage customers, Sobel said. Evolve's customers typically use the provider’s cloud storage services to retrieve a backed up virtual machine, but Sobel said his firm can now recreate a customer’s environment in the event of a disaster.
“We could start spinning up a customer’s virtual environment from backups stored at our facility within 30 minutes of a disaster event," he said. "Particularly for small to mid-sized businesses, the cloud could serve as their hot site.”