How do you ensure I/O write order when performing fast copy backups for z/Linux under z/VM?
Our environment: - z/VM 5.3 - z/Linux SuSE SLES 9 - DB/2 9.0 - Oracle 10g - EMC DASD with TimeFinder SNAP capability As a secondary means to recovery data in event of a local disaster, all DASD is dump to tape and move offsite. The process consists of four steps. 1. All z/Linux servers and the z/VM system are shutdown. 2. Using the EMC SNAP feature (or FlashCopy if it was IBM DASD) with the consistency option, the DASD is fast replicated to target volumes. 3. Several z/OS ADRDSSU jobs are used to dump the volumes to tape. 4. While the z/OS jobs dump the target volumes, z/VM and the z/Linux servers are started. Since we can perform a consistency snap, that is a point in time copy of the source volumes, we're considering leaving VM and the Linux servers up while the snap occurs. My concern has to do with the recovery of the DB/2 and Oracle databases. I'm unsure of how the z/Linux and z/VM caching may affect write order sequence. If the database and the database logs are on different file systems, is it possible the writes may not be written to the DASD in the same order as they were issued? Are there options to ensure the write order is maintain? By the way, we use mirrored DASD to a remote site as our primary means to recover from a disaster. We had several DR tests where VM and Linux servers were recovered successfully. However, the applications that run on the Linux servers have low transaction rates so I expect there not much I/O occurring. Arthur T. Hallock U.S. Office of Personnel Management 202-606-1195 ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
