I suppose nothing "special" except that you'll have very few options
for allocating resources between the two database instances on each
node in the cluster. This is why I personally strive for a single RAC
instance on each node of the clusters I deal with. In most cases,
people are using RAC for their most sensitive, mission-critical
application, so it's usually relatively easy to make the argument for
finding somewhere else to put the "other" databases. This usually
provokes a discussion about whether the database(s) in question
*really* are mission-critical or not. Or, alternatively, consolidate
them into a single database where some resource allocation can be made.
This problem is really not much different than putting multiple single-instance databases on a standalone, non-clustered server. Dan Andrew Kerber wrote: -- //www.freelists.org/webpage/oracle-l |