This discussion is archived
8 Replies Latest reply: Sep 28, 2011 7:25 AM by 755353 RSS

primary stand alone - standby 2 node physical rac?

865330 Newbie
Currently Being Moderated
Hi,

is it possible to have primary as stand alone database and standby as 2 node physical rac?

are there any extra steps that i need to perform while creating standby db ?
  • 1. Re: primary stand alone - standby 2 node physical rac?
    CKPT Guru
    Currently Being Moderated
    is it possible to have primary as stand alone database and standby as 2 node physical rac?
    are there any extra steps that i need to perform while creating standby db ?
    First RAC means, you will have single redo thread and also single undo thread for each instance,
    If your primary database has only one thread then how can you build standby on RAC?

    However good question.
  • 2. Re: primary stand alone - standby 2 node physical rac?
    Dr. Paranoid Pro
    Currently Being Moderated
    Yes you can. Many customers have done this to move to RAC with very little downtime. We do not have specific documentation on it but the steps are generally as follows;

    GOAL:  Use Data Guard to execute the migration from a single instance production system to a new RAC platform with no more than 10 minutes of total database downtime.  In addition to reducing downtime, this process mitigates business risk by having a functioning standby database that is always readily available to roll back to, should unforeseen problems be encountered. 

    Source system - current Production system
    Target system - new RAC system.

    Phase 1: Preinstall all RAC software on the Target System. Create a single node Physical standby database with cluster database set to false.  The Source system is the primary production system.   The detailed steps are provided in the Data Guard documentation for instantiating a standby database.  

    Phase 2: Enable redo shipping using maximum performance mode.  As precautions to avoid a situation where the primary can stall if there are not enough arch processes available to archive on-line logs use the following two parameters on the Primary database.  

    Set max arch processes higher than the default of 2.  This will prevent all arch processes from being consumed by the process of catching up the standby database with archive logs generated since the backup was taken.  

    Also, be sure to configure the Source system to enable it to function as a standby that will receive redo from the Target when role transitions are tested in Phase III below.  This is also discussed in the Data Guard documentation under Role Management.

    Phase 3:  Testing.  Once the standby is caught up with the primary it is possible to stop apply (archive logs continue to be shipped) and open the database read only to test.  To catch up again, simply restart apply.

    If possible, also test switchover.  Reverse roles during slow periods, changing the standby to primary and back again.  This will give timings and confidence for the final migration.   This will also require some downtime of the Primary database so plan accordingly.

    Phase 4:  Enable the RAC cluster on the Target system and convert the physical standby to a RAC database.  This can be done with Enterprise Manager if the database is at least 11.1.0.7 or higher using the Covert to RAC wizard.

    Phase 5:  Do the final switchover to "promote" the Target system to the primary production role.  Maintain the original Source database as a Physical standby database to have an up-to-date system that can be quickly rolled back to should there be an issue with the new primary system.  

    Phase 6:  Continue use of the Source system as a single node (reduced failover capacity) or covert it to a RAC as well to create a symmetrical Data Guard.

    Larry  
  • 3. Re: primary stand alone - standby 2 node physical rac?
    CKPT Guru
    Currently Being Moderated
    @Larry,

    Thanks for your great explanation,

    Here i have some confusion.
    Phase 4: Enable the RAC cluster on the Target system and convert the physical standby to a RAC database. This can be done with Enterprise Manager if the
    database is at least 11.1.0.7 or higher using the Covert to RAC wizard.

    To convert RAC as you said we can use EM or RCONFIG, fine. After performing from single instance to RAC database.

    How the archives will applied on standby, as primary has only one thread, and standby database(cluster) has two instances, So how it coordinate to apply the archvies on each instance?

    Thanks.
  • 4. Re: primary stand alone - standby 2 node physical rac?
    865330 Newbie
    Currently Being Moderated
    CKPT wrote:
    @Larry,

    Thanks for your great explanation,

    Here i have some confusion.
    Phase 4: Enable the RAC cluster on the Target system and convert the physical standby to a RAC database. This can be done with Enterprise Manager if the
    database is at least 11.1.0.7 or higher using the Covert to RAC wizard.

    To convert RAC as you said we can use EM or RCONFIG, fine. After performing from single instance to RAC database.

    How the archives will applied on standby, as primary has only one thread, and standby database(cluster) has two instances, So how it coordinate to apply the archvies on each instance?

    Thanks.
    Hi CKPT

    if the standby is RAC then only one instance can apply the redo at any given time.
  • 6. Re: primary stand alone - standby 2 node physical rac?
    865330 Newbie
    Currently Being Moderated
    Phase 4: Enable the RAC cluster on the Target system and convert the physical standby to a RAC database. This can be done with Enterprise Manager if the


    I have my standby created as standalone now, if I fail over to standby and then convert it into RAC , will my current primary also become RAC database? this is physical dataguard i assume all data dictionary stuff are the same ...
    or will I have to recreate the current primary after I failvoer to the standby and convert the standby into RAC ?

    do I need to do performance test of the system after this stage ?
  • 7. Re: primary stand alone - standby 2 node physical rac?
    BPeasland Guru
    Currently Being Moderated
    How the archives will applied on standby, as primary has only one thread, and standby database(cluster) has two instances, So how it coordinate to apply the archvies on each instance?
    A Standby database in RAC will have 1 node in managed recovery and the other nodes in the cluster are in MOUNT mode sitting there basically idle. The one node applying recovery can easily apply 1 thread, 2 threads, or more. If the primary has N nodes, the standby doing the apply must wait for all N threads to arrive at the standby before it can do the apply. In this case, it only has to apply the 1 thread.


    HTH,
    Brian

Legend

  • Correct Answers - 10 points
  • Helpful Answers - 5 points