This article can also be found in the Premium Editorial Download "Storage magazine: Optimizing your enterprise database storage."
Download it now to read this article plus other related content.
The Enterprise Storage Group takes an even dimmer view of backup success rates, estimating that 60% of backups do not properly execute in networked storage environments.
Why the poor results? In a word, human error, says Schrenk. "The guys forget to put in the tapes," or perhaps "they don't set a long enough backup window," he offers.
Furthermore, a lot of the time, storage administrators aren't even aware that a backup failed - especially in multiserver environments. Checking on the status of each and every server getting backed up is sometimes easier said than done. Vyant's WorldView addresses this issue by installing an agent on each catalog server that reports back "to the mothership"-a central database with a Web-based console - every 15 minutes, says Schrenk. If a backup fails, administrators can drill down into the software's error codes.
Further complicating matters are mixed backup environments-not only heterogeneous hardware, but different backup packages. To that end, Vyant's WorldView as well as BackupReport from Bocada gather information from a variety of backup packages, be they Veritas NetBackup, Legato Networker, HP OmniBack, or CA BrightStor.
At Reader's Digest, the Global
"It was a pretty sorry state," recalls desktop technology analyst Adam Ladd, whose group took over Veritas backup operations for about 100 Unix, Novell and Windows servers. "Backups were failing, and we had no time to rerun them," he recalls. "We had no sense of how much we were growing, or how long jobs were taking to complete." In lieu of tools, the group wrote scripts and used Viso to create project timelines in the form of Gantt charts to predict and plan backups.
Installing BackupReport changed all that. With the insight Reader's Digest gathered from BackupReport, the group decided to start multistreaming its backups, mix the jobs throughout the day and uncovered unexpected areas of growth. "We're over the hump," says Ladd. These days, "a backup fails maybe once a week - if that."
This was first published in November 2002