Data replication service

Edit online

The Data Replication Service (DRS) is an internal, proprietary component of the WebSphere® Application Server. The DRS is used by other components to move data from node to node within a network of clustered application servers.

Figure 1. Moving a cache entry across cluster members using DRS
This figure shows how DRS is used to share data between the four cluster members, 1 through 4.

In Figure 1, DRS is used to share data between the four cluster members, numbered 1 through 4.

Figure 2 illustrates a WebSphere cluster. The workload is balanced between two IBM® HTTP servers, which take input from all four WebSphere cluster members.
Figure 2. Example of a clustered WebSphere topology
The following charts reflects a clustered WebSphere topology.

The Data Replication Service caters to two scenarios: failover and caching. This study analyzes the caching feature; failover scenarios were not analyzed.

When an object is available in the cache, repeated requests for the same information are handled faster. Because the data in the cache must be the same, irrespective of which application server the request arrived at, it makes sense to cache the same data in of all the application servers. Data Replication Service handles moving cache entries from server to server.

Figure 3. Topology configurations
This figure shows the topology required for V6 Distributed Replication Services (DRS) and DynaCache DRS to function properly.

In the default topology, (left side of Figure 3), each server in the domain holds a replica of the data from one other server. In the second example, (right side of Figure 3), the double headed arrows mean that data flows from each process to every other process, so that for each replicated object, there are three remote copies and the local original. The type of topology is specified when the replication domain is created. Then, the entire domain selection corresponds to this topology.

This topology would probably be more than what is needed for HTTP session replication. In the DynaCache environment, the topology described in the right side of Figure 3, is the only allowable configuration for cache replication because when caching dynamic content, the cache is only useful if it is available on all the machines where a request could arrive.

This study used only the caching capability with the topology as shown in the right side of Figure 3.