Following the September 11th terrorist attacks, the storage networking industry found itself in an awkward situation. Disaster recovery has always been one of the solutions that SAN advocates have promoted. Use of Dense Wave Division Multiplexing (DWDM) or new IP storage technologies to extend the reach of storage beyond primary data centers has enabled remote mirroring and remote backup of vital corporate data. Fortunately for the collective karma of the storage networking industry, few vendors have attempted to opportunistically exploit the sudden surge of interest in disaster recovery. Instead, most have simply restated the technology benefits that SANs generally provide for disaster recovery, highlighting new functionality that can enhance design and implementation of business continuance strategies.
Like tape backup in general, disaster recovery has not been a favored career path for storage administrators. Safeguarding ever-changing corporate data is often a monotonous routine that, like paying insurance premiums, is done grudgingly but on a fixed schedule. While backing up vital corporate data to tape has been assisted by the development of larger tape libraries and higher capacity tape cartridges, the primary issue for tape backup is not scheduling or tape rotation, but restoring data from tape in the event of failure. Restoration imposes a lengthy downtime during which ongoing business may be lost. Consequently, despite the higher cost of disk mirroring to maintain a readily accessible copy of data, mirroring strategies are increasingly deployed as part of an optimum disaster recovery solution.
Intelligent controllers in higher end storage devices offer new options for disk mirroring, including the ability to have two separate storage units use each other as secondary mirrors. For disaster recovery implementations, this would allow two separate data centers to provide mutual support in the event of the failure of either center, performing on a grand scale the failover capability of server clustering. Mirroring between two remote sites, however, is not a trivial undertaking, particularly if long distances are involved. Depending on the applications being run to storage, the amount of data that must be synchronized between two locations may require sustained gigabit bandwidth in both directions. Initial attempts to use (DWDM) for extending distance were successful within a metropolitan circumference, but have not scaled well to regional distances. Mirroring between New York and New Jersey, for example, can be done with simple SAN extension via DWDM. Mirroring between New York and Chicago is problematic.
Unfortunately for all of us, the distances now required to truly safeguard human and corporate data resources are unknown. Where previously it might have been sufficient to mirror between two buildings in the same city, now an entire city may be victim to direct or indirect disruption from an attack. Consequently, consultants are now recommending business continuance solutions that span regions, maintaining mirrors that may be separated by hundreds or thousands of miles.
The precedent for long distance mirroring was established by businesses situated in geologically unstable areas, such as California or the Pacific Northwest. Until recently, achieving stable, high performance links between regions has been difficult. Carriers such as Qwest are now offering gigabit and multi-gigabit services that can span coast to coast. For storage applications such as remote mirroring, new IP storage technologies have demonstrated sustained gigabit performance over thousands of miles. These new solutions are enabling disaster recovery strategies no longer bounded by distance or speed considerations, and which can now concentrate on the more critical issues of preserving human and IT resources in the event of catastrophe.