In a message dated 9/13/2005 4:21:48 P.M. Central Daylight Time, [EMAIL PROTECTED] writes:
>YMMV. Can back up a mod-3 in a couple of milliseconds. The "back up" does not really complete in a couple of milliseconds. What happens is that the controller makes a bit map of all the tracks you want to be FlashCopied during those milliseconds. From then on the controller knows which tracks to allow updates to and which not. The target volume will be guaranteed to have a point-in-time consistent copy of your source tracks. After doing the one I/O to establish the FlashCopy relationship, some other work is necessary. You need to allocate a new data set on the target volume covering the tracks you just "backed up" on the source. Then you must run a backup job whose input is the target volume's data set. A read to a target track that has not been changed gets redirected to its corresponding source track, and a read to a changed track goes to the target volume, because the controller copied the source track before allowing the write to take place to that source track. This backup job may take hours, depending on the amount of data to be copied, the interference from other I/O activity, and other things that affect mileage. But the backup copy is all internally consistent as of the fateful moment when those couple of milliseconds went flying by. Bill Fairchild ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
