Timing for Cloning with Flashback Database vs other methods

From: Bellows, Bambi \(Comsys\) <bbel5_at_allstate.com>
Date: Fri, 5 Dec 2008 11:42:55 -0600
Message-ID: <AD0CB572A820AB4E8E52ABD38950FD3605803EF4@a0001-xpo0150-s.hodc.ad.allstate.com>

Hey there Listers!  

Cloning sure has more options nowadays than in the golden yesteryear, doesn't it? There's the simple file copy with the controlfile you backed up from trace... being the oldtimers' best friend; then there's the newfangled "duplicate database" , which has the oldtimers smackin their gums talkin about walkin to school in the snow, 5 miles each way, uphill, without boots. And, as if that weren't enough, there's this snazzy glittery clone achieved through flashback database, which has the oldtimers sittin in the dark waitin fer the young dbas to change the lightbulbs. And fine. But. Which one's fastest? We oldtimers could parallelize that copy and shove it to background faster than you could say gollygeewillikers... can RMAN do that? And what of this snazzydazzy flashback database for cloning? Can it start with a blank slate, or does it need a database to be created and stuff before it lets tear? And, really, seriously, what are the timings? One would assume that if the target database exists and is pretty much kept up to date, anything which just applies changes is going to be lickety-split faster, but what if it's not?  

I'm off to soak my teeth for awhile, but I appreciate any insights you whippersnappers might have... :-)


Received on Fri Dec 05 2008 - 11:42:55 CST

Original text of this message