All of the TIBCO BusinessEvents® Extreme high availability features can be used across a WAN to support application deployment topologies that require geographic redundancy without any additional hardware or software. The same transactional guarantees are provided to nodes communicating over a WAN, as are provided over a LAN.
Figure 7.15, “Geographic redundancy” shows an example system configuration that replicates partitions across the WAN so that separate data centers can take over should one completely fail. This example system configuration defines:
Partition
Awith node listOne,Two,FourPartition
Bwith node listThree,Four,Two
Under normal operation partition A's active node
is One, and highly available objects are replicated to
node Two, and across the WAN to node
Four, and partition B's active node
is Three, and highly available objects are replicated
to node Four, and across the WAN to node
Two. In the case of a Data Center
North outage, partition A will transition to
being active on node Four in Data Center
South. In the case of a Data Center South
outage, partition B will transition to being active on
node Two in Data Center
North.
The following should be considered when deploying geographically redundant application nodes:
network latency between locations. This network latency will impact the transaction latency for every partitioned object modification in partitions that span the WAN.
total network bandwidth between locations. The network bandwidth must be able to sustain the total throughput of all of the simultaneous transactions at each location that require replication across the WAN.
Geographically distributed nodes should be configured to use the static discovery protocol described in the section called “Location discovery”.