Suggestions for Using Tuning Parameters

There are several basic recommendations for improving Replication Server performance. Whether or not changing these configuration values improves your system performance depends on your system configuration and how Replication Server is used at your site.

• Set the Amount of Time SQM Writer Waits
  Use the init_sqm_write_delay and init_sqm_write_max_delay Replication Server configuration parameters to set the amount of time SQM writer waits.
• Cache System Tables
  Use the sts_cache_size and sts_full_cache_table_name Replication Server configuration parameters to cache system tables.
• Executor Command Cache
  Use the Executor command cache to cache column names and datatypes for a primary Adaptive Server database table, when a Sybase RepAgent initially sends an insert, delete, or update LTL command for that table.
• Stable Queue Cache
  Replication Server uses a simple caching mechanism to optimize I/O. This mechanism reduces write latency and improves reader speed, since data can usually be read quickly from the cache.
• Asynchronous Parser, ASCII Packing, and Direct Command Replication
  Obtain improvements throughout the replication process during data transformation and transport by utilizing the asynchronous parser, ASCII packing, and inbound and outbound direct command replication features together.
• Incremental Parsing
  Incremental parsing improves replication performance especially for replication environments with multiple Replication Servers in the replication route, and reduces memory consumption for HVAR, RTL, and DSI bulk-copy-in.
• Command Conversion
  Use command conversion to alter the replication of certain commands either for all or for selected replicated tables.
• Set Wake up Intervals
  Use the rec_daemon_sleep_time and sub_daemon_sleep_time Replication Server configuration parameters to set wake up intervals.
• Size the SQT Cache
  Use the sqt_max_cache_size Replication Server configuration parameter and the dsi_sqt_max_cache_size database connection configuration parameter to size the SQT cache.
• Control the Number of Outstanding Bytes
  Use the exec_nrm_request_limit, exec_sqm_write_request_limit, and md_sqm_write_request_limit database connection configuration parameters to control the number of outstanding bytes of memory.
• Control the Number of Network Operations
  Use the dsi_cmd_batch_size database connection configuration parameter to control the size of a DSI command batch.
• Control the Number of Commands the RepAgent Executor Can Process
  Use the exec_cmds_per_timeslice database connection configuration parameter to control the number of commands the RepAgent executor thread can process.
• Specify the Number of Stable Queue Segments Allocated
  Use the sqm_recover_segs Replication Server configuration parameter to specify the number of stable queue segments Replication Server allocates before updating the RSSD with recovery QID information.
• Select Disk Partitions for Stable Queues
  Use the disk_affinity database connection configuration parameter to specify the logical name of the partition to which the next segment should be allocated when the current partition is full.
• Make SMP More Effective
  Use the smp_enable Replication Server configuration parameter to enable symmetric multiprocessing (SMP).
• Specify the Number of Transactions in a Group
  You can use different configuration parameters to control the number of transactions in a group.
• Set Transaction Size
  For single DSI connections, set the value of dsi_large_xact_size to the maximum value of 2,147,483,647. Even when parallel DSI is not configured, the DSI/S reads the statement limit set by dsi_large_xact_size and performs several tasks related to parallel DSI.
• Enable Nonblocking Commit
  Use the dsi_non_blocking_commit Replication Server configuration parameter to enable nonblocking commit by specifying the number of minutes to extend the period of time Replication Server saves messages after a commit.
• Memory Consumption Controls
  Replication Server can show warning messages when the memory consumption exceeds a defined threshold, and you can control the memory used by the EXEC, DSI, and SQT threads.