Heterogeneous Multi-Path Replication

Use multiple replication paths to increase replication throughput and performance, and reduce contention.

In a single-path replication environment, transactions replicate serially from the primary database to the replicate database to ensure the primary database transaction commit order, and therefore to ensure that the replicate database is consistent with the primary database. The serial mode of applying transactions to the replicate database remains, even though multiple applications typically execute their respective transactions in parallel at the primary database, or even if there are transactions arriving from multiple primary databases.

There are replication environments that can maintain data consistency within a subset of tables, without serializing all transactions that originate from the same primary database. A typical example of this environment is when different applications that access different sets of data modify a single primary database. The different sets of data within the subset of tables that are modified by a specific application continue to replicate serially. Data in different subsets of tables can replicate in parallel.

Multi-Path Replication™ supports the replication of data through different streams, while still maintaining data consistency within a path, but not adhering to the commit order across different paths.

A replication path encompasses all the components and modules between the Replication Server and the primary or replicate database. In multipath replication, you can create multiple primary replication paths for multiple Replication Agent connections from a primary database to one or more Replication Servers, and multiple replicate paths from one or more Replication Servers to the replicate database. You can configure multi-path replication in warm standby and multisite availability (MSA) environments. You can convey transactions over dedicated routes between Replication Servers to avoid congestion on shared routes, and you can dedicate an end-to-end replication path from the primary database through Replication Servers to the replicate database, to objects such as tables and stored procedures.

See Replication Server > Administration Guide Volume 2 > Performance Tuning > Multi-Path Replication to configure Adaptive Server as the primary or replicate database in a multipath replication system.

• Parallel Transaction Streams
  Multi-path replication can improve replication performance as long as transactions can be divided into parallel streams, and do not have to be serially committed across different streams.
• Default and Alternate Connections
  In multipath replication, connections include the default and one or more alternate connections.
• Interfaces File Requirements for Sybase IQ
  Create an entry for the Replication Server in the interfaces file (sql.ini file in Windows) of the Sybase IQ data server. If there are multiple Replication Servers replicating to a single Sybase IQ server, you must add all Replication Servers to the interfaces file.
• Dedicated Routes
  A dedicated route distributes only transactions for a specific primary connection. You can create a dedicated route to the replicate Replication Server to replicate high priority transactions or to maintain a less congested path for a specific primary connection.
• Heterogeneous Multi-Path Replication Scenarios
  Use the scenarios to build multiple replication paths between Adaptive Server, Oracle, and Sybase IQ databases.