Simplifying disaster recovery solutions to protect your data - Disaster Recovery

Computer Technology Review, Dec, 2003 by Ravi Chalaka

Implementing a remote data replication policy is the first step toward a comprehensive disaster recovery plan, which is no longer an option but a necessity for many IT computing environments. The government has recognized the enormous risk represented by the loss of critical data and is stepping up its efforts to mandate certain requirements across industries such as healthcare, banking, brokerage and insurance. For example, all healthcare-related industries were required to comply with the U.S. Department of Health and Human Services' Health Insurance and Portability Act of 1996 (HIPPA) by the end of 2003. Among the many mandates of this act are data availability requirements to ensure that information is available at all times.

In response to the recovery deficiencies that surfaced in the aftermath of the September 11th terrorist attacks, the SEC has published new disaster recovery requirements for banks and brokerage houses mandating both off-site requirements and recovery time limits. The Enron fiasco resulted in the Sarbanes-Oxley act that imposes new records retention standards for corporate governance.

Customers are struggling to comply with these new government requirements, but deployments can be stifled due to technological or budgetary constraints of the previously available solutions based on proprietary products, which demand expensive lock-ins to one storage vendor. Other solutions are difficult to manage or are unable to leverage existing infrastructures, imposing further complexity and cost barriers.

Mirroring Clarified

In addition to the basic replication process, data mirroring can also be employed as part of the disaster recovery (DR) solution. The process of mirroring involves the use of a "shadow" disk that is updated in parallel with the primary disk, providing a real-time or near real-time copy of the primary disk. Local mirroring provides the first level of data protection with a mirror disk attached to the host machine or an appliance located at the primary site. In the event of data loss on the primary disk, the data is retrieved seamlessly from the mirror disk.

There are two types of mirroring techniques that may be used over a remote IP to establish a copy of the primary site data--synchronous or asynchronous. Synchronous mirroring solutions' performance depends on the link bandwidth (speed) and distance spanned by the remote link. Each write transaction to the disk is sent to the remote mirror disk and the application cannot continue until that transmission is acknowledged from the remote location. Synchronous mirroring solutions provide maximum data protection at the expense of degraded primary site performance and reduced network throughput on the link to the DR center. An optimum alternative that has minimal impact on performance while minimizing data loss is asynchronous mirroring. With this technique, multiple writes are transmitted without waiting for individual acknowledgments. The use of asynchronous mirroring offers the additional benefit of "near real-time" availability of data; online standby of data is only a few writes behind the primary site.

Data Replication Made Easy

Data replication is the basis of all disaster recovery solutions and involves periodically copying a volume's data onto a secondary storage device, which can be located any distance from the original--preferably far away. If the main storage device should fail, data on the secondary storage device can immediately be promoted to primary status and brought online. Replication is a continuous process that begins by establishing a complete copy of data at risk at the disaster recovery (secondary) site. With that copy as a baseline, the replication process continues, recording any changes to data and forwarding those changes to update the secondary site based on watermarks with a user-specified policy.

Fast Recovery

For true protection against major disasters, DR centers require remote sites to be located hundreds of miles away from the business, raising issues of data loss and synchronization of data between the production and DR site. The ideal disaster recovery solution provides quick time-to-recovery (TTR), assuring continuance of operations (near uninterrupted access to data). Additionally, the solution must minimize data loss for a graceful recovery, keeping primary and DR center data synchronized while minimizing human intervention to reduce errors during the recovery process.

For robust data protection, the remote replication solution must also work in conjunction with options for multiple point-in-time snapshots to provide full, incremental, or differential automated instant backup to disk. This allows administrators not to have to spend needless hours trying to recover accidentally deleted records or virus-infected files from tape, even in the event of a disaster.

Snapshot agents minimize recovery time. When data replication is used in conjunction with snapshot agents, the data has full transactional integrity in addition to point-in-time consistency. This means the replica can be immediately put into active use without going through a complete 'consistency check' process that can be very time-consuming for large databases. Simply put, DR and replication solutions require flexibility, ease of deployment, scalability and rapid recovery.