SQL Server Performance

Replication data corruption

Discussion in 'Performance Tuning for SQL Server Replication' started by cchitanu, Jul 25, 2006.

  1. cchitanu New Member

    Here is my problem : from time to time the data on a table used in replication is getting corrupted at the subscriber(transactional replication, push subscription) : character like Ó is copied as O.

    Both publisher and subscriber database, tables have the same collation : SQL_Latin1_CP850_CI_AS, however subscription server has a different collation : SQL_Latin1_General_CP1_CI_AS. A fix we found : if we delete and re-insert the row at the publisher this time the data is transferred fine by the replication.

    Any idea ?

  2. dineshasanka Moderator

    Is there any possibility of hacing sql collation for both publisher and subscriber


  3. satya Moderator

    Can you make sure to keep the Database Collation is similar between all the servers, if not you can mention the collation on the table wide and column wide to manage the replication.

    Satya SKJ
    Microsoft SQL Server MVP
    Contributing Editor & Forums Moderator
    This posting is provided �AS IS� with no rights for the sake of knowledge sharing.
  4. cchitanu New Member

    Hi Satya,
    Collation at database and table level is the same. We can't force the same collation at the server level because the subscriber has multiple publishers with different collations.
    Another observation is that the collation used are : SQL_Latin1_General_CP1_CI_AI(case insensitive and accent insensitive) so is there any chance that this collation setup is responsible for "data corruption" ?


Share This Page