Alejandro Vargas' Blog

Resizing the online redologs on a RAC environment is a simple task, that require a little more attention than that of a single instance.

Usually we do resize when redolog switches happen too frequently. Take care to add all the new larger groups, then switch over until the current redolog for each instance is on the new larger groups before starting to remove the old groups.

You can find a step by step description of the procedure on this document: How to Resize the Online Redologs on a RAC Environment

In some cases it may happen that you have only one instance of the RAC database up:

SQL> select inst_id,thread#,status,enabled,groups,instance,sequence# from gv$thread
/

   INST_ID    THREAD# STATUS ENABLED      GROUPS INSTANCE    SEQUENCE#
---------- ---------- ------ -------- ---------- ---------- ----------
         1          1 CLOSED PUBLIC            3 sati1              41
         1          2 OPEN   PUBLIC            3 sati2              58

You will still see that both threads have a current redolog with status active:

SQL> select inst_id,group#,thread#,sequence#,bytes,members,status from gv$log
  2  /

   INST_ID     GROUP#    THREAD#  SEQUENCE#      BYTES    MEMBERS STATUS
---------- ---------- ---------- ---------- ---------- ---------- ----------------
         1          1          1          0   10485760          2 UNUSED
         1          2          1          0   10485760          2 UNUSED
         1          3          1          0   10485760          2 UNUSED
         1          4          2         59   10485760          2 CURRENT
         1          5          2          0   10485760          2 UNUSED
         1          6          2          0   10485760          2 UNUSED
         1         10          1         39   52428800          2 INACTIVE
         1         11          2         56   52428800          2 INACTIVE
         1         12          1         40   52428800          2 INACTIVE
         1         13          2         57   52428800          2 INACTIVE
         1         14          1         41   52428800          2 CURRENT
         1         15          2         58   52428800          2 INACTIVE

In this case you will not be able to remove the current group of the thread belonging to the instance that is down until you start the instance and switch over this thread to the new groups; if you try to remove the current group of a down instance you will get the following error:

SQL> alter database drop logfile group 14;
alter database drop logfile group 14
*
ERROR at line 1:
ORA-01623: log 14 is current log for instance sati1 (thread 1) - cannot drop
ORA-00312: online log 14 thread 1: '+DATADG/sati/onlinelog/group_14.303.648232897'
ORA-00312: online log 14 thread 1: '+DATADG/sati/onlinelog/group_14.289.648232903'

Extended RAC is a Tier 6 DRP technology that enable zero data loss in case of major disaster at a local site, and near 0 downtime if well configured and well managed.

It is an expensive technology because of the costs associated with the required infrastructure, dark fiber over long distances and redundant components on all hardware layers.

A strict change control procedure must be in place, with an identical test environment where all changes to the infrastructure and new code versions are carefully tested before implementation on the production site.

Oracle 11g provides new features that greatly help on implementing extended clusters, and makes it the version that best fit the challenge.

The syncronization between sites is obtained using ASM redundancy, reads are made locally at each site using ASM preferred mirror read feature, and in case of a local storage failure ASM fast mirrot resync make recovery much easier.
   
10g R2 and 11g Dataguard FSFO (Fast Start Failover) technology provide a similar level of protection at less cost and at greater distances, as it relays on TCP for communications, so it is convenient to consider it as an option when evaluating extended RAC for DRP.

This document: Extended RAC On Pure Oracle Stack Summary contains basic implementation guidelines.

More information can be found on these excellent documents: