Just a second...

Multi-datacenter replication

The multi-datacenter capability replicates client sessions and the R/ branch of your topic tree in a different data center as part of your Diffusion™ Cloud service.

Note: Multi-datacenter support is only enabled on the Pilot, Launch, Scale, and Dream editions of Diffusion Cloud. For more information, see https://www.pushtechnology.com/pricing/.

If your Diffusion Cloud service fails, the replicated data is used to recover your service from another datacenter within seconds.

The replicated information includes the following items:
The replicated information does not include the following items:
  • Any routing topics included in the subset of the topic tree. For more information, see Topic replication
  • All topics not included in the subset of the topic tree. For more information, see Topic replication
  • Console configuration (if this add-on is enabled).

All of your usage and billing information is unaffected by the failure and recovery of your Diffusion Cloud service.

Session replication

If the your Diffusion Cloud service becomes unavailable, your clients experience a brief disconnection. When the clients attempt to reconnect on the same URL, they are routed to the recovered service. If the clients are configured to reconnect within 60 seconds, they reconnect to the same session, which has been replicated.

The following session information is replicated:
  • session ID
  • session principal
  • session properties
  • list of topic selections

Topic replication

The replicated data includes a copy of all topics in the R branch of the topic tree – for example, R/foo/bar. Replicating these topics generates no additional messages.

The following topic types can be replicated:
  • JSON
  • Binary
  • Single value
  • Record
  • Stateless
  • Slave

    If the master topic that the slave topic refers to is not in the replicated branch of the topic tree, the slave topic is put in a pending set. The slave topic is not present on the recovered service until the master topic is re-created, at which point the slave topic is automatically created on the recovered service.

The following topic information is replicated:
  • The structure of the R branch of the topic tree. Any topics that make up the structure of this branch but are not one of the supported types listed are not created and the topic definition and data are not replicated. Instead the topic path of these unsupported topics remains an empty tree node.
  • The topic definitions for all topics in this branch. For example, the topic type and any metadata associated with the topic.
  • The topic data for all topics in this branch.

When a client that subscribes to a topic in the R branch of the topic tree reconnects to the recovered service, it continues to receive updates from the topic.

Replicated topics must be updated by exclusive updates. To update a topic in the R branch of the topic tree, a client must register as an update source for that topic.

When a client that is the active update source for a topic in the R branch of the topic tree reconnects to the recovered service, it must re-register as a topic update source to continue to update that topic.