No, I wasn't suggesting that you were implying Linux NFS is kosher for outside of the playground - I was more commenting that the overall DBA mentality of avoiding Linux NFS at all costs is not without historical merit. And of course, I know you work for Polyserve and certainly have the chops to talk about Linux NFS (we know some of the same people). Thanks, Matt -----Original Message----- From: oracle-l-bounce@xxxxxxxxxxxxx [mailto:oracle-l-bounce@xxxxxxxxxxxxx] On Behalf Of Kevin Closson Sent: Wednesday, April 05, 2006 4:46 PM To: oracle-l@xxxxxxxxxxxxx Subject: RE: 10gR2 RAC on single SLES9 node >>>In the interest of fairness, however, we once tried to deploy our >>>reasonably large Oracle RAC development environment on Linux NFS >>>servers as a cost-saving measure when we were a Wee Company (tm). We >>>discovered some wonkiness in the Linux NFS server code would cause Oh boy. Where in the world did I ever get misconstrued as suggesting simple-stupid-plain-jane-vanilla-Linux servers exporting NFS mounts as an option for "reasonably large RAC development environment" ? The topic at hand is how to set up a multi-instance, single Server RAC test/training kit where I say use NFS on loopback for simplicity. Of course taking some braindead vanilla Linux FS and exporting it for NFS mounting is NOT A SOLUTION for anything in the database arena. There is a HUGE difference between NAS and NFS Server exports. A NAS device (e.g., NetApp,EMC Cellera, HP EFS-CG) has a lot of specialized NFS server-side stuff that makes it hold together for things like databases or hard-hitting file serving. Since PolyServe is funding the V4 NFS development at CITI, there is an ever-so-slight chance that it will get even better. Disclaimer. I work for PolyServe and I have intimate knowledge about what databases on NAS are supposed to look like. -- //www.freelists.org/webpage/oracle-l -- //www.freelists.org/webpage/oracle-l