A replication topology is a defined arrangement between independent Netmail Store clusters, connected to one another via Content Router nodes. Content Router supports several alternative replication topologies.
On this page:
Content Router allows an administrator to replicate some or all the streams stored in a primary cluster to a disaster recovery (DR) site. In case of a complete failure or loss of the primary cluster, all replicated streams can be recovered from the DR site. At the discretion of the administrator of the primary cluster, a subset of streams may be replicated to the DR.
Content Router allows mirroring between two or more primary clusters. All designated streams stored in Cluster A will be replicated to Cluster B, and vice versa. The administrator of each cluster can decide, based on stream metadata, which streams should be replicated to the other cluster. It is possible that the two replication sets will be exactly the same set of streams, or completely disjoint sets, or overlap in some other way.
Multi-Site Disaster Recovery
Using a combination of one-way DR replication and mirrored clusters, more complex disaster recovery topologies can be deployed. Such deployments rely upon a unique metadata identifier for the origin of each object to ensure the correct subset of content is recovered from a pooled (many-to-one) DR cluster should a disaster occur. For additional information about metadata, see Content Metadata.
At the discretion of the administrator of the primary cluster, the subset of streams replicated to each DR site need not be identical. Some or all the DR sites may be shared with other independent primary clusters. In case of loss of the primary cluster, all replicated streams can be recovered from one or more DR clusters. This requires modification of the rules set in the DR cluster, and it is recommended that you seek advice from Messaging Architects Professional Services before attempting this.
Below is a figure showing multiple primary clusters rolling up to one DR cluster. In order to distinguish the streams in Primary Cluster 1 from those in Primary Cluster 2, metadata needs to be stored with each stream identifying, at a minimum, the cluster of origin.
An alternative use for Content Router replication infrastructure is to roll up or distribute content within an organization or between cooperating organizations. With a little forethought and planning when storing descriptive metadata with each stream, a very sophisticated data distribution and storage infrastructure can be created. Because you have the ability to create your own rules, a dynamic pool of data can be created and moved around with relative ease. In the example below, a combination of mirrored and one-way replication is used to create a network of clusters with distinctly different data sets. The only two identical clusters are the Central Repository in Singapore and the DR Cluster in Utah. This same model could be used for roll up of major functional areas, such as Financial, Legal, Engineering, HR, etc.