CIO — What happens when a telephone company loses its customer billing data, or when an online brokerage firm's performance degrades so badly that stock-trading transactions are delayed? What happens when a retail store's point-of-sale system goes down, or when customers looking for bank loans online find that the bank's system is down?

In today's Internet economy there is little loyalty, and customers will readily defect if their needs for high availability, reliability and performance are not met. When mainframe data is lost, corrupted or cannot be accessed quickly, it can be catastrophic. People typically associate "threats to business continuity" with disasters such as hurricanes, floods or terrorist attacks. These calamitous events can wipe out entire data centers, severely disrupting business.

But other types of events, such as human errors, application errors, and delays in reacting to changing conditions in the mainframe environment can also disastrously disrupt business continuity. Human errors can wipe out critical data. Application errors can stuff erroneous data into business-critical databases. A delay in responding to a change, such as a spike in workload, can drag down the performance of a business-critical application. According to leading industry analysts, as much as 80 percent of all unplanned downtime is caused by software problems or human error. Companies must be prepared to deal with these types of events.

Driving Factors

Gone are the days when the mainframe was monolithic, walled off from the outside world and accessed only by a small number of skilled IT personnel. Today, the mainframe is a vital and tightly integrated resource in the enterprise IT infrastructure. As part of a complex, multitiered, services-oriented architecture, the mainframe must interoperate with a variety of other resources. For example, a single SAP landscape can include mainframes, multiple servers and hundreds or thousands of database tables.

Additionally, the mainframe must meet today's demand for 24/7 operation. Maintenance windows have virtually disappeared, forcing operations staff to perform maintenance tasks—such as deploying bug fixes, and upgrading and adding new applications—while the system is operational or in the very limited downtime window for those tasks that cannot be done while the systems are up and running.

What's more, today's Internet environment has opened up the mainframe to thousands, even millions, of outside users—employees, business partners and customers. As a vital component of business processes, the mainframe must participate in business-to-business transactions with systems outside the enterprise, such as those in supply and distribution chains.

The resulting complexity has increased the potential for human error and for errant code in applications. It increases the likelihood that people will make operational errors that can cause data loss. Complexity also increases the probability of coding errors in applications, which can result in the contamination of business-critical databases. For example, in updating a database, an administrator makes a single keystroke error in a batch update job that causes the update to be performed with the wrong input data set, contaminating critical business data.

Continue Reading