Oracle RAC has always had a latency issue when the RAC nodes are
geographically distributed. In these cases, many shops use Oracle
Streams from back-and-forth replication. RAC on Extended Distance
Clusters seeks to eliminate those latency issues by providing fast
(sub-minute) recovery from a site failure using Dark Fiber (DWDM or
CDM) technology.
 |
We have RAC
experts who specialize in RAC architecture. Just call
for expert RAC failover consulting, including distributed
RAC systems. |
In this architecture, all nodes at all sites are allowed to be
part of a single database cluster. Due to the inherent advantages
provided by RAC on Extended Distance Clusters, interest in the
architecture has been on the rise, however it is important to know
what situations Extended Distance Clusters are ideal for, especially
considering the impact of distance and latency, as well as the
complexity of the implementation.
The practical limitations for RAC on Extended Distance Clusters
stem from the high impact of latency, and thus distance. The ideal
environment for implementing this architecture is where the
datacenters are less than 25km apart, and where direct cables with
dedicated channels between the sites have already been set up. Due to
the impact of distance on performance, it is unlikely that RAC on
Extended Distance Clusters will be implemented on any distances
greater than 25km
