3 Replies Latest reply on Feb 2, 2013 6:32 PM by budachst

    Way to make a mirror in repository (LVM, Clone?)

      Dear all,

      A customer have 2 pillar axiom 600, and present of each pillar one 3 TB LUN to OVM server (2 servers in cluster). They want to make a way to mirror the LUN of storage 1 in Storage 2 via OVM, like a LVM Mirror. I'm thinking in the best way to accomplish this:
      The clone function will make a real time replication, or one single point-in-time clone?
      Can I make a mirror in Server OS with LVM commands and create a repository with this logical volume?
      Someone have a suggestion of a better way do to this?

        • 1. Re: Way to make a mirror in repository (LVM, Clone?)
          Its called "replication" and most all storage can perform this function. Might cost some money to license it but it works well.

          Clone is only a single point in time clone. I imagine you can use the CLI and schedule it to run every so often. You can only hot clone from a FC or iSCSI LUN. Not NFS. Even then, the target will have to be checked common file system errors. True replication software on the storage side will not have this issue. Bottom line, you shouldn't use Oracle VM to do "replication" of storage repositories.
          • 2. Re: Way to make a mirror in repository (LVM, Clone?)
            Yes, I replicate the storage inter sites (I have 2 sites, one is failover of other) with maxrep, but in each site i have 2 storages.
            I will try to use OVM_CLI to make schedule clones of my virtual disks to repository in storage 2. Thanks!
            • 3. Re: Way to make a mirror in repository (LVM, Clone?)
              OVM itself doesn't provide anything of this low-level volume manager-like functions. You will either have to replicate your LUNs using the storage solutions you already have, or come up with some other means of mirroring.

              E.g. I am using two ASM nodes that provide a multipathed iSCSI target to the same ADVM volume, which is mounted on my OVM servers and then formatted using OCFS2. In the backend, the ADVM volume(s) are made up from three storage arrays (3 ASM faliure groups) per ASM disk group. This provides quite some redundancy to my storage repos.