How Remote Copy and Replication Are Different from Backup

On the surface, remote copy, replication, and backup seem to be the same things. In all three cases, data is copied from a primary storage device, such as a disk array, to a secondary one. They all sit in the same layer of the information model (Figure 4-0). The goals are the same, too: to safeguard data by making a copy of it.

Figure 4-0. Remote copy and replication reside in the data management layer

There would also appear to be similarities between remote copy, replication, and RAID. RAID copies data to multiple disks so that a hot copy is available at all times. This protects against total system failure due to a disk failure. It also provides protection in the form of a second copy in case the drive failure renders the original data unrecoverable. Remote copy does compare closely with RAID, but on a grander scale.

Backup protects data by moving it to another location. RAID protects data by continuously copying an entire disk's worth of data so that every change is recorded. Remote copy and replication do both. With remote copy and replication, data is copied to a safe location continuously or nearly so. This ensures that complete, up-to-the-minute copies of data are safe and immediately available for failover in the event of disaster.

What is distinct about remote copy and replication is the scope. Backups and RAID are designed to guard against local failures and disasters the loss of a disk or array. Remote copy and replication guard against wholesale, regional destruction, including natural disasters and terrorist attacks. It allows data to be tucked away in a safe place, awaiting widespread disruptions in telecommunications, electricity, buildings, and other essential infrastructure.

Another important aspect of remote copy and replication that differs from backup is its quick recovery time. With backups, a new storage device may need to be purchased or installed and then the backup data placed on it. Even for the fastest backup systems, this can be a time-consuming operation. With remote copy or replication, an exact duplicate already exists on another storage device that can be accessed immediately. Even complete failure of a system may not be noticed by end-users if the system fails over to the secondary array immediately.

How Far Is Far Enough?

A key question when considering remote data copy techniques is how far away the data should be placed from the main facility. It depends on the threat against which you want to guard. There are numerous local threats that affect only a single building or two, such as fire, water-main breaks, and local electrical failure. To have a second copy of the data in a bunkerlike disaster facility on the other side of town is enough.

Natural disasters are different. Earthquakes, hurricanes, and tornadoes can destroy buildings in a wide area and disrupt communications and electricity in entire regions. Safeguarding against data loss from natural disasters means placing the secondary data in a different region, perhaps hundreds of miles away.

Some horrible disasters cannot be dealt with effectively. Nuclear war, for example, would bring destruction over such a wide area that it is nearly impossible to guard against data loss fully and be operational in anything approaching a normal fashion. Only governments should worry about this. Corporate enterprises should focus on those things that they can deal with regional and local disaster. Analyze the threat to the data, and determine distance from that.