You may think that your organization’s disaster recovery solution is full-proof, but you’re probably wrong. In fact, even synchronous replication cannot ensure zero data loss and rapid recovery in a rolling disaster scenario.
A rolling disaster usually starts with a power outage or a problem with network communication lines. It is an event in which replication to the secondary (disaster recovery) site fails before the primary (production) data center fails. During the interim between replication failure and failure of the primary data center, the primary data center’s production servers continue to produce data that is not being replicated (unless you have Axxana’s Phoenix). Suffice it to say, when data is created and not replicated—even for a few seconds—data loss is guaranteed.
These Assumptions Will Take You Down
Despite your best-laid plans and steep investments in traditional disaster recovery solutions, your organization may be in for data loss and prolonged, costly recovery times if you make the following assumptions:
- Synchronous Is Fail-Safe – Synchronous replication is designed for disasters in which the production data center fails before (or at the same time as) the lines used to replicate data to the backup data center (or a DRaaS provider’s data center). In rolling disasters, where replication lines fail first, your production data center continues to produce data, which you will lose and cannot recover.
- Max Availability = Zero Data Loss – Maximum Availability mode is commonly perceived as synchronous replication in an Oracle® Data Guard environment, but it cannot ensure zero data loss in rolling disasters. This is because Data Guard cannot operate without fully functioning replication lines. When these lines fail in a rolling disaster, your application(s) will continue to operate without replicating the data to the disaster recovery site.
- Asynchronous Data Lag Is Controllable – If you’re running asynchronous replication, your organization is probably willing to tolerate a certain amount of data loss. But rolling disasters may create a much larger volume of yet-to-be replicated data—and therefore greater data loss—than your organization has planned for or can withstand. For example, what happens if replication fails during a period of peak throughput? At this point, data loss may move from being tolerable to becoming a liability.
- Three DCs Is the Best That Money Can Buy – In a three-data-center topology, your organization replicates synchronously to a nearby data center and asynchronously to a more distant data center. If a rolling disaster terminates synchronous replication to the nearby data center, and the primary site continues to produce data, your organization will lose that data. In other words, in spite of its tremendous complexity and cost, your three-data-center solution will still fail you.
In the Midst of Chaos, Every Disaster Must Be Treated Like a Rolling Disaster
To make matters worse for disaster recovery and business continuity planning, the exact chain of events leading to a disaster isn’t always obvious. If you can’t determine whether you’ve had a rolling disaster, you need to respond and remediate as if you have. Doing so requires time-consuming identification of what data has been lost, and in some cases, reconstruction of application consistency at the secondary site. Meanwhile, availability, productivity, revenue, and reputation may slide downhill.
No Organization Is Immune
If you think your organization is immune to rolling disasters because power and network outages rarely occur or won’t affect you, consider this: In a recent survey, respondents listed power outages and network outages as the first- and second-leading cause (respectively) of failover to a secondary site.
Don’t say we didn’t warn you.
For More Information
To find out more about Axxana’s Phoenix and how it guarantees zero data loss at any distance and preserves data consistency for rapid recovery—all at a vastly lower cost than traditional disaster recovery solutions, read our white paper on rolling disasters or contact
 J. Buffington et al. Enterprise Strategy Group. Research Report: The Evolving Business Continuity and Disaster Recovery Landscape. January 2016. http://www.cioandleader.com/zerto/pdf/Research-Report-The-Evolving-Business-Continuity-and-Disaster-Recovery-Landscape.pdf