Since I’m on site at a Beta customer (testing the product I work on at Oracle), this will be a quick blog entry. I’ve been meaning to direct folks to Gear6 for quite some time now. I have no stake in Gear6, so this is not a shameless plug. I think they solve interesting problems so if you are a large NFS shop, I’d recommend checking them out. They offer a plug-in NFS read-through cache and while I haven’t had first hand experience with their product, I know folks that have and they had good things to say about Gear6.
If any of you are confused about what NFS has to do with Oracle, I recommend this list of Oracle on NFS related posts.
Kevin Closson's Blog: Platforms, Databases and Storage 
I am in the middle of a POC with Gear6/NetApp 6070 and a 3 node Oracle RAC RHEL4 cluster now, with early positive results.
bw.
Hi MR Welty,
Is this an 11g or 10g PoC? I’m very interested!
Hello sir,
– We are already started on a project implementing Oracle 11g RAC on 2 highend Solaris 10 update 4 (SPARC) servers.
– We have no intention of using any 3rd party clustering software (either for storage management or application)
– Oracle binaries will be installed on the local disk on each of the 2 servers
– I had assumed all other files will go on ASM (raw devices via FCP on NetApp 6070), but the current requirement is that control files, redo logs and the archive logs stored on an actual shared filesystem (for ease of recovery for example).
Data files will still go on the ASM disks using raw devices via FCP.
This means we either use a cluster file system (OCFS2 would have been OK if it were available for Solaris) or as I suggested
– NFS (Oracle DirectNFS). These files will be shared between all the instances and the filesystem will be mounted concurrently for read/write on both servers. So if one server “dies”, the other can recover it’s services since all necessary files are accessible by both nodes
Questions:
1. please what are your thoughts on this.
2. Does it make sense?
3. Any hints, guides available?
Kevin,
To the comment from MrWelty, I worked with Welty on this POC and it was an Oracle 10G RAC with three RAC servers. The caching of the Oracle database was able to off load the NetApp by 45-50%. This POC worked great, after a some bumpiness, of course! 🙂