As Published In March/April 2004

HA: How Affordable?
By Jeff Spicer

Availability is critical, and now it's not as costly and complex.

System downtime is expensive—very expensive, depending on your business. In some industries when a critical system or application fails or is taken offline, the cost can run millions of dollars per hour—and that's just in lost sales or transactions and doesn't take into account damage to customer relationships and the future health of the business. And the longer the downtime lasts, the more damage that's done. According to a recent article in the Disaster Recovery Journal, 50 percent of companies that lose critical business systems for more than 10 days never recover.

Years ago, the demands placed on systems and applications were different. Many organizations had the luxury of planned system downtime—intentionally taking systems, applications, or other resources offline for maintenance or upgrades. A business could afford, for example, to take its order-entry system offline over a weekend for maintenance because all of its customers were also closed for business. But now, international businesses "follow the sun," operating offices around the globe during local normal business hours; manufacturing lines run round the clock; Web-based businesses and online interfaces allow consumers and enterprises to place orders any time day or night; and the list goes on. Many of those underlying systems, applications, and resources that once could be periodically offline are now critical and must run 24/7.

But this continuous operation of systems and applications is only part of the high-availability (HA) equation, the other major piece being the reduction of unplanned downtime caused by accidents, equipment failure, and natural disasters. Imagine how the failure of a disk, for example, might result in the loss of critical data. Or the loss of electrical power might cause a financial application to go offline during quarter-end number crunching.

Until recently, ensuring HA was a costly and complex endeavor. A rough rule of thumb used to be that making an application highly available could triple the cost of deploying that application. And each 9 added to the measure of that application's necessary availability (99.9 percent versus 99.99 percent, for example) added yet more cost. But Oracle and its partners have been working to bring the cost of HA to affordable levels. Oracle Real Application Clusters (RAC), for example, provides a straightforward method of building a clustered system that can automatically handle hardware failure. And RAC technology is designed to run on low-cost commodity hardware, helping control costs associated with HA.

Oracle has offered an array of high-availability technologies for years, such as Oracle Data Guard, the management infrastructure for creating and monitoring standby databases; Recovery Manager for backing up and recovering your Oracle database; LogMiner for analyzing DML updates and DDL changes in the database and rolling them back if necessary; Flashback technologies for point-in-time recoveries or simply to query past versions of schema objects; failover clusters and Real Application Clusters; and Oracle Streams and Transportable Tablespaces for keeping large amounts of data synchronized.

Next Steps LEARN more about Oracle HA technologies VISIT the OTN grid technology center

And now with Oracle 10g's focus on enterprise grid computing—the pooling of servers and storage into flexible, on-demand resources that can be used by the applications that need them—high availability is even more deeply woven into the fabric of the database. And because enterprise grid computing relies on innovations such as low-cost blade servers, open-source operating systems like Linux, and small multiprocessor servers, availability is rapidly becoming even more affordable.

Oracle 10g also offers new technologies organizations can use when creating and managing highly available systems. For example, a new feature called Automatic Storage Management (ASM) allows the database to manage the underlying storage. The benefits for HA are many: the database can now redistribute data across a storage pool when a new disk is added, helping cut back planned downtime; ASM helps manage database I/O to keep it optimal; and ASM even provides automatic mirroring.

In this issue of Oracle Magazine, author David Kelly interviews businesses using Oracle technologies to build highly available systems. Read their stories and discover how these forward-thinking companies are using Oracle technologies to build affordable, highly available systems.