Any data center manager can tell you that his or her company's data stores are growing at staggering and exponential levels. In fact, analysts say the total amount of digital information in the world is now measured in zettabytes (one million petabytes)! As if that were not enough, we live and work in a global environment across time zones and geographies.
The traditional "backup window", when an organization can stop working to back up their data, no longer exists. This means the existing model for backup and storage management is no longer sufficient, and the paradigm has shifted from managing and executing backups in a nightly window, to protecting and managing data on an ongoing basis. In addition, the backup process has transformed from a tape-based model to a disk-based model, with many abandoning tape altogether and utilizing disk-to-disk (D2D) with off-campus replication for disaster recovery safety. In order to eliminate the backup window and solve the data growth issue what is needed is to rethink how we protect data altogether.
A redesign is necessary whereby we move from copying data and storing it into a proprietary backup set to a comprehensive data protection solution that provides holistic server mirroring with near-instant recovery. Mirroring net new changes every 5 or 10 minutes eliminates the backup window, brings recovery point objectives and recovery time objectives (RTO/RPO) to near-zero, reduces or eliminates tape costs, helps reduce backup software licensing costs, and provides local and remote disaster recovery (DR). The side benefit, is that having your servers cloned in their native format allows for you to run test/dev and DR validations off your out of band recovery environment without effecting production. A solution such as this provides workload (application) mobility and sets forth the ability to easily test and migrate applications to any data center whether internally owned, a colocation facility or even a public cloud. This capability also provides the ability to failover or migrate entire data centers in hours and not days.