Introduction to Data Management

8 Protecting Databases
With RDF
15873 Tandem Computers Incorporated 8-1
Sometimes, a truly major disaster occurs at a computer site. The disaster might be a
fire, an earthquake, flooding from a broken water pipe, or release of a toxic substance
that severely damages the computer hardware. Catastrophes like this happen rarely.
But when they do, they can take down the entire computer system—and perhaps even
destroy all systems at the site, leaving the site with no way to carry on its daily
business operations.
In the conventional world of batch processing, recovery from this kind of disaster
customarily requires much time and effort. It can be a complex, labor-intensive, and
error-prone process that sometimes takes as long as two days to complete. This
cumbersome process results in a resource outage that most OLTP applications simply
cannot tolerate. For these applications, a system shutdown of just a few hours can
result in a substantial business loss.
OLTP customers with computer systems at multiple sites want to use these
distributed resources to safeguard their applications and data if a disaster strikes at a
particular location. They would like an arrangement where an alternate site takes
over for the damaged site in a matter of seconds or minutes, with minimal impact on
transactions in progress and the database. Through the Remote Duplicate Database
Facility (RDF), Tandem provides exactly this kind of protection.
RDF monitors TMF transaction activity on a database residing on a designated
primary system and applies duplicate updates to an identical copy of the database
residing on a remotely located backup system in the same network. The backup
database is continually updated by RDF, providing a current, online copy of the
primary database, as illustrated in Figure 8-1. The backup database can be located
nearby or across the nation. Thus, RDF can protect against area-wide or even regional
disasters.