Zero data loss is the ideal, the aspiration. Is it achievable? That is a question that exercises the minds of service providers in the cloud or otherwise.

Most cloud service providers have two or three data center “replication solution”. This solution advocates that data should be replicated redundantly to duplicate the production environment and reduce the time to recover. A failover recovery is initiated when the primary data center servers fail suddenly and the second or third data center smoothly takes over as the primary data center.

A few cloud service providers tweak their software to permit the creation of local copies of data synchronous with creation of data backups on a remote server. The argument is that time to recover can be reduced if local copies of the data is available to the end consumer for recovery to physical servers post disaster. The remote server copies of the data can be used to perform bare metal recoveries of information if the disaster wipes out the data center and the local copy of the data.

Other cloud service providers encourage their customers to use continuous backup schedules to back up their data to the remote server. Thereafter, the service provider offers to replicate the data on to a removable media for ship back to the customer.

However, the type of data center selected by the enterprise will be determinant on the value of Zero Recovery Point Objective (RPO) that has been set by the organization. If the real time order entry will impact the business, a zero RPO is automatic. If there is a margin of tolerance, a near zero time RPO may be selected and that RPO may also be relaxed.

It follows that the selection of a Zero data loss disaster recovery environment will require the answers to some of the basic questions. These questions must be asked and answered before the zero data loss environment is selected or created.

  1. What is the revenue contribution of the protected application?
  2. What is the storage capacity that needs to be protected?
  3. What is the performance latency that can be tolerated?
  4. What is the I/O rate for the applications that must be protected?
  5. What will be the business impact of an outage?
  6. What is the possibility of an outage?
  7. What is the cost of primary storage?