+

Search Tips   |   Advanced Search

Multi-broker replication domain settings

Use this page to configure a multi-broker replication domain. This console page applies only to replication domains that were created with a previous version of the product. Replication domains use the data replication service (DRS). To view this console page, click Environment > Replication domains > multibroker_replication_domain_name.

An application server that is connected to a replicator within a domain can access the same set of data sent out by any application server connected to any other replicator, including the same replicator. Data is not shared across replication domains.


Name

Name for the replication domain. The name must be unique within the cell.


Request timeout

Number of seconds that a replication domain consumer waits when requesting information from another replication domain consumer before giving up and assuming that the information does not exist. The default is 5 seconds.

Information Value
Data type Integer
Units Seconds
Default 5


DRS partition size

Number of groups into which a replication domain is partitioned. By default, data sent by an application server process to a replication domain is transferred to all other application server processes connected to that replication domain. To filter or reduce the number of destinations for the data being sent, partition the replication domain. There should be at least one server listening to every partition. If there are no servers listening on a partition, all the replicas created in that partition are lost because there is no server to cache the objects. The default partition size is 10, and the partition size should be 10 or more to enhance performance.

Partitioning the replication domain is only applicable if you are replicating data to support retrieval of an HTTP session if the process maintaining the HTTP session fails. Partitioning is not supported for sharing of cached data maintained by web container dynamic caching. As to dynamic caching, all partitions or groups are always active and used for data replication.

When you partition a replication domain, you define the total number of groups or partitions. Use this setting to define the number of groups. Then, when you configure a specific session manager under a web container or as part of an enterprise application or web module, select the partition to which that session manager instance listens and from which it accepts data. To specify the groups to which an application server listens, change the settings for affected servers on a session manager page. In addition, we can set a role or runtime mode for a server. This role or mode affects whether an application server process sends data to the replication domain, receives data, or does both. The default is both to receive and send data.

Information Value
Data type Integer
Default 10


Single replica

Specifies that a single replication of data is made. Use this option only if you are using session manager with memory to memory replication. Enable this option if you are replicating data to support retrieval of an HTTP session if the process maintaining the HTTP session fails. This option restricts the recipient of the data to a single instance.

Avoid trouble: Do not enable this option on a domain that is using dynamic cache replication.gotcha This setting provides filtering beyond grouping or partitioning. Using this setting, we can choose to have data only sent to one other listening instance in the replication domain.

Information Value
Default false


Serialization method

Specifies the object serialization method to use when replicating data. An administrative concern with replicating Java objects is locating the class definition, especially in a Java EE environment where class definitions might exist only in certain web modules or enterprise applications. Object serialization methods define whether the processes receiving data also need the class definition.

The options for this setting are OBJECT and BYTES. The default is BYTES.

OBJECT instructs a replicator to write the object directly to the stream. With OBJECT, a replicator must instantiate the object on the receiving side so it must have the class definition.

BYTES instructs a replicator to break down the object into bytes and then send only the bytes across the stream. With BYTES, a replicator does not need to instantiate the object on the receiving side. The BYTES option is useful for failover, where the data is not used at the receiving side and the class definitions do not need to be stored on the receiving side. Or, the option requires that you move class definitions from the web application class path to the system class path.


DRS pool size

Size of the pool of resources allocated for communication with its Java Message Service (JMS) transport. We must configure this number to be the same as the DRS partition size. The default is 10.

Pooling replication resources can enhance the performance of the internal data replication service.


DRS pool connections

The domain replication service should create a pool of connections with its Java Message Service (JMS) transport rather than reusing a single connection. We can pool connections when using a single replica or client server environment. You should not pool connections in a peer to peer environment.

The default is to not create a pool of connections for replication.


Subtopics


Related concepts

  • Multi-broker replication domains


    Related tasks

  • Replicating data with a multi-broker replication domain

    Replication domain collection

    Replicator entry collection