Disaster Recovery

Provision for Disaster Recovery is often a fine balance between what is required by businesses and what is affordable. Ideally all IT infrastructure should be fully duplicated and data should be constantly replicated off-site, so as to cater for any level of disaster, ranging from the simple loss of a file to a complete and total site disaster. The reality is however that very few organisations are prepared or required to duplicate costs to this extent, and that a balanced approach could normally be obtained. The following are some of the key elements to be considered in building a disaster recovery plan:

Backups. These are really not negotiable to any organisation and daily, weekly, monthly and yearly backups are an absolute must.

Do we need immediate server recoverability? If this is the case, then the server clustering is really the only option. This is offered through a number of different options, including MSCS, VERITAS Clustering and VMware’s VI3. This essentially provides a physical or logical standby server available at all times, allowing applications to be taken over within minutes.

Do we need to be able to rebuild a server image? This may be a cheaper way of overcoming the problem of recovering from a broken/ failed server. Rather than clustering or keeping a standby server/ virtual machine available, an image of the entire system is taken and stored on disk, also on a SAN. With the server imaging technology available today, this has become a trustworthy process and may be a good alternative.

Do we need to replicate data off-site? Having servers recovered in an environment where all data and disks have been destroyed may be of little value. Tape backups will get data restored, but the time involved may be too long. Much as data replication (and in particular synchronous data replication) requires significant bandwidth, it may at times be the only viable and acceptable solution. Vendors like Symantec has successfully intergraded CPS (continuous protection server) technology with backups, making for a complete and integrated solution.

These are just a few of the issues that need to be considered, with a cost qualification in the context of each one of them.

Too often IT departments carry the responsibility for setting requirements for DR rather than the business units who it affects most. For this reason, organisations often engage Perfekt to talk with the various application owners, to create an independent document of requirements. This is then presented in the format of a gap analysis between what is required and what is in place at the time, enabling our customers to pursue a structured and planned path to a business-wide acceptable solution.