Suggested Deployment Model for Disaster Recovery

To set up disaster recovery, a suggested model is to have a primary grid in one location and a mirror or disaster recovery (DR) grid in another location. This provides redundancy when the entire location hosting the primary grid experiences a disaster and requires failover to another location.

Grids in a gridset do not need to have the same number of copysets. They should be sized so that their capacity is sufficient to take over in the event of a disaster.

Both control and data traffic flow between each of the copyset nodes and the proxies configured in remote grids. All nodes in all copysets must be able to contact the proxies configured for DR in all other grids in the gridset.

The diagram shown later in this topic illustrates the deployment model. In this case, the primary data grid comprises a primary realm service and three copysets (each containing three replicas) behind a firewall. Data is then being mirrored across a WAN link (the cloud shape) to another location where there is another firewall and then a proxy. The mirror grid on the other location comprises a satellite realm service and a group of two copysets (each containing three replicas).

This demonstrates the ability of a mirror grid that has a different number of copysets from the primary grid. The dashed lines in the diagram also show the data flow where the control traffic and data traffic are sent to the proxy at the mirror grid, which is what you must configure with specific IP addresses and ports so that the proxy at the mirror grid is accessible to the primary grid. In addition, the proxy at the primary grid must be accessible to the mirror grid for traffic to flow in that direction as well. In the event of a disaster when the primary grid location becomes inaccessible, you can manually set the mirror grid to be the new primary grid as of the last consistent checkpoint that was mirrored to that location.

Figure 2: Deployment Model