4 Replies Latest reply: May 14, 2014 1:15 AM by 902923 RSS

    GG on RAC

    902923

      Hi Experts,

       

      We have a replication requirement between oracle to oracle database. This is four node rac configuration, here we have a doubt , lets assume

       

      GoldenGate has installed and configured on  both side and extract created with threads 4 option,  user connected node 1 and there is a dml transaction  happening in source database   , before extract process fetching the committed data from node 1-online redo log file the node got down , in this scenario  How the missing committed data from node 1 log file will be handled by GoldenGate without data loss ? how that data will be captured by ogg if particular log file node is down? please clarify

       

       

      Thanks in Advance.

        • 1. Re: GG on RAC
          Frak!!!

          Hi,

          Normally Oracle GoldenGate use DB login information to connect to the database, so even one node is down as long as the connection of OGG user is still exists, the GoldenGate can retrieve the commited data from the redo log file using the survival nodes.

          • 2. Re: GG on RAC
            902923

            Thank Tareq,

             

            Can you please explain ,is there any different fetching functionality in OGG while using FAILOVER_MODE = (TYPE = SELECT)(METHOD = BASIC)) and FAILOVER_MODE = (TYPE = SESSION)(METHOD = BASIC)) in Oracle RAC configuration ?

             

            For Example,  Normally if there is an node failure in RAC while transaction happening, Oracle request / response process will take care of that failuer node commited transactions like roll backward method and helped to write the data in to data files, like that how gg will capture the roll back process data if node failed ?Please clarify.

            Thanks in Advance.

            • 3. Re: GG on RAC
              Nick.W-Oracle

              If the redo logs are in ASM, OGG will continue to use the ASM connection to pull the details out of the redo log.   If the redo logs are on raw devices, they should be shared to all systems, so even if the node goes down, the online log will be available for reading.   The only event where it would not be available is if the storage server crashed that held the online redo logs or the ASM files... in that case, there isn't much OGG can do because the system is down.

               

              Once it comes back up, OGG can be restarted and it will pick up right where it left off and continue processing all the transactions from all nodes.

               

              Another option, is to use Downstream Integrated Extract.

               

              Nick

              • 4. Re: GG on RAC
                902923

                Thanks Nick.