Data replication domains

Data replication domains

Data replication domains and multi-broker domains both perform the same function, which is to provide data replication between application servers in a cluster. Even though we can still configure existing multi-broker domains with the current version of the product, after we upgrade the deployment manager, we can only create data replication domains in the administrative console.

For transitioning users: A replication domain created with a previous version of the product might be a multi-broker domain. We should migrate these multi-broker domains to data replication domains. Data replication domains allow us to:

Configure all of the instances of replication that need to communicate in the same replication domain.
Configure the session manager with both types of replication domains to use topologies such as peer-to-peer, and client-to-server to isolate the function of creating and storing replicas on separate application servers.
Control the redundancy of replication for each type of replication domain, because with a data replication domain, we can specify a specific number of replicas.

If we used multi-broker domains with earlier releases of the product, use the following comparison chart to learn the differences between how v5.x and v6.x and v7.0 application servers use the two types of replication domains:

v5.x application servers using replication domains V6.x and v7.0 application servers using replication domains
Replication domain types Uses only multi-broker replication domains for replication. Servers that are using the current version of the product can be configured to use both multi-broker replication domains and data replication domains for replication. The two types of domains provide backward compatibility with multi-broker domains that were created with a Version 5.x server. We should migrate any multi-broker domains to data replication domains.
Data transport method Uses multi-broker domain objects that contain configuration information for the internal Java Message Service (JMS) provider, which uses JMS brokers as replicators. Uses data replication domain objects that contain configuration information to configure the high availability framework on the product. The transport is no longer based on the JMS API. Therefore, no replicators and no JMS brokers exist. We do not have to perform the complex task of configuring local, remote, and alternate replicators. The earlier version of the product did not support data replication domains. The current version of the product can be configured to perform replication using old multi-broker domains by ignoring any JMS-specific configuration and using the other parameters to configure replication through the high availability framework.
Replication domain configuration The earlier version of the product encourages the sharing of replication domains between different consumers, such as session manager and dynamic cache. The current version of the product encourages creating a separate replication domain for each consumer. For example, create one replication domain for session manager and another replication domain for dynamic cache. The only situation where we should configure one replication domain is when configuring session manager replication and stateful session bean failover. Using one replication domain in this case ensures that the backup state information of HTTP sessions and stateful session beans are on the same application servers.
Partial partitioning We can configure partial partitioning. Partition the replication domain to filter the number of processes to send data. Partial partitioning is deprecated. When using data replication domains, we can specify a specific number of replicas for each entry. However, if we specify a number of replicas larger than the number of backup application servers running, the number of replicas is the number of application servers running. After the number of application servers increases above your configured number of replicas, the number of replicas created is equal to the number specified.
Domain sharing Multiple data replication service (DRS) instances share multi-broker domains. A limitation exists on the number of multi-broker domains that we can create because every multi-broker domain contains at least one replicator. A maximum of one replicator can be on each application server. All DRS instances in a replication domain use the same mode. Each replication domain must contain either client only and server only instances, or client and server instances only. For example, if one instance is configured to client and server, all other instances must be client and server. If one instance in a replication domain is configured to be a client only, we can add client only and server only instances, but not a client and server instance.

Related:

Data replication
Replicating data across application servers in a cluster

	v5.x application servers using replication domains	V6.x and v7.0 application servers using replication domains
Replication domain types	Uses only multi-broker replication domains for replication.	Servers that are using the current version of the product can be configured to use both multi-broker replication domains and data replication domains for replication. The two types of domains provide backward compatibility with multi-broker domains that were created with a Version 5.x server. We should migrate any multi-broker domains to data replication domains.
Data transport method	Uses multi-broker domain objects that contain configuration information for the internal Java Message Service (JMS) provider, which uses JMS brokers as replicators.	Uses data replication domain objects that contain configuration information to configure the high availability framework on the product. The transport is no longer based on the JMS API. Therefore, no replicators and no JMS brokers exist. We do not have to perform the complex task of configuring local, remote, and alternate replicators. The earlier version of the product did not support data replication domains. The current version of the product can be configured to perform replication using old multi-broker domains by ignoring any JMS-specific configuration and using the other parameters to configure replication through the high availability framework.
Replication domain configuration	The earlier version of the product encourages the sharing of replication domains between different consumers, such as session manager and dynamic cache.	The current version of the product encourages creating a separate replication domain for each consumer. For example, create one replication domain for session manager and another replication domain for dynamic cache. The only situation where we should configure one replication domain is when configuring session manager replication and stateful session bean failover. Using one replication domain in this case ensures that the backup state information of HTTP sessions and stateful session beans are on the same application servers.
Partial partitioning	We can configure partial partitioning. Partition the replication domain to filter the number of processes to send data.	Partial partitioning is deprecated. When using data replication domains, we can specify a specific number of replicas for each entry. However, if we specify a number of replicas larger than the number of backup application servers running, the number of replicas is the number of application servers running. After the number of application servers increases above your configured number of replicas, the number of replicas created is equal to the number specified.
Domain sharing	Multiple data replication service (DRS) instances share multi-broker domains. A limitation exists on the number of multi-broker domains that we can create because every multi-broker domain contains at least one replicator. A maximum of one replicator can be on each application server.	All DRS instances in a replication domain use the same mode. Each replication domain must contain either client only and server only instances, or client and server instances only. For example, if one instance is configured to client and server, all other instances must be client and server. If one instance in a replication domain is configured to be a client only, we can add client only and server only instances, but not a client and server instance.