Nearly 90% of Oracle users experienced significant downtime last year due to both planned and unplanned events,
despite the fact that the vast majority say they require continuous uptime, a new survey of Independent Oracle Users Group (IOUG) members found.
The survey of 265 IOUG members, including database administrators (DBAs), developers and managers, also found that most companies have not yet implemented clustering or grid technology in order to ensure higher levels of availability.
IOUG president Ari Kaplan said that what struck him most about the survey results was the gap between what enterprises are prepared for and what they need. Most respondents cited a necessity for constant uptime, but "57% had no business continuity plan," Kaplan said. "That's a big discrepancy."
Kaplan said another surprising find was that most unplanned interruptions to database access were not caused by external events such as natural disasters or security breaches. "The number-one cause was old-fashioned hardware failures," Kaplan said. The next most common reasons were network failures, database errors or bugs, application errors and power glitches, he added.
In addition to unexpected problems, a number of members reported system downtime due to planned occurrences -- 63% had to take down their databases to perform upgrades of hardware and 56% interrupted availability for routine maintenance. Database and applications upgrades were also common causes of planned downtime.
"If the window for planned maintenance is larger than the window you can be down, you really have to come up with a strategy to deal with it," Kaplan said. "There's a cost involved there, but your business has to weigh how much it costs against how [important it is that you] be up 24x7 all year."
The types of companies that require continuous availability include large firms in the financial services, government and health care fields, Kaplan said. Companies with world wide Web-based operations also need uninterrupted uptime. Kaplan explained that for an e-commerce site such as Amazon, every minute of downtime hurts your business as well as your image, because the media reports on it.
Implementing a high-availability (HA) solution such as Oracle Grid or Real Application Clusters (RAC) could reduce the incidence of applications going down. Kaplan said these technologies provide extra server power to take on the burden in the case of a single point of failure.
With these Oracle products or a third-party HA tool "you reduce the top unplanned outages and the vast majority of issues," Kaplan said. "The good news is you can plan for it."
Currently, less than a third of respondents have adopted RAC, and about 13% use Oracle Grid. Another 20% and 24% are considering RAC and grid, respectively.
Most businesses continue to rely on tape and disk-based backup as a continuity strategy, according to the survey. However, it can take 48 hours or more to bring systems back up with this method following an event.
Kaplan emphasized the need for workers to align their continuity plan with their business requirements. "It's pretty telling that businesses are saying that [they] need to be operating more than they are planning for," he said.