Forum Stats

  • 3,875,196 Users
  • 2,266,853 Discussions
  • 7,912,106 Comments

Discussions

GG - Recover from Network Failure

837734
837734 Member Posts: 49
edited Feb 28, 2011 7:15AM in GoldenGate
Hi
I have one Oracle Database Server. Golden Gate is used to replicate data to another Oracle Database Server. If there is any network connection failure between Source and Target Database Servers then what are ideal steps to be followed once the network connection is established to restart replication ?

Thanks
Hiren Pandya

Best Answer

  • stevencallan
    stevencallan Member Posts: 3,459
    edited Feb 28, 2011 5:58AM Answer ✓
    The checkpoint table and file, with GoldenGate using its CSN, will do that for you. If you have the log files (using that to encompass both archived redo logs and current online log files), yes. As an experiment, stop the data pump to simulate loss of network connectivity. Do some DML on the source in addition to some log switches. Turn data pump back on. Check the target lag, and then when EOF for lag, you're caught up.

Answers

  • stevencallan
    stevencallan Member Posts: 3,459
    Recover from network failure - Which is the entire point of using a secondary extract process (data pump), so that trails are created and loss of network connectivity does not become a single point of failure in your replication environment. How long can you tolerate a network outage is related to how long you are going to keep redo logs around. If the appropriate logs are available, GG will resume from the point of failure. Otherwise, you need to make the log files available. If they are gone, you need to resync or start from a later point in time and deal with the data loss separately.

    The "ideal" steps are to use a data pump and dodge the loss of redo in the first place. Explicitly stated quite a few times in the admin guide.
  • 837734
    837734 Member Posts: 49
    Hi Steven
    Thanks for the reply. I am using Data Pump and the database server is configured in Archive Log Mode.

    I did not understand your question : "how long you are going to keep redo logs around." . Redo Log Files and Archive Log Files will be availalbe 24 * 7.

    So is it that Golden Gate will automatically start reading appropriate log files after network connection is established ?

    Thanks for your suggestions.

    Hiren Pandya
  • stevencallan
    stevencallan Member Posts: 3,459
    edited Feb 28, 2011 5:58AM Answer ✓
    The checkpoint table and file, with GoldenGate using its CSN, will do that for you. If you have the log files (using that to encompass both archived redo logs and current online log files), yes. As an experiment, stop the data pump to simulate loss of network connectivity. Do some DML on the source in addition to some log switches. Turn data pump back on. Check the target lag, and then when EOF for lag, you're caught up.
  • 837734
    837734 Member Posts: 49
    Thank you so much Steven. I shall test it.
This discussion has been closed.