Paolo wrote:
Also, I am not sure if there are others who need to host multiple datasets
on the same machine... but, lately, I have been wondering if it would make
sense to share the node table and related caches between different TDB
datasets hosted on the same machine and running in the same JVM.
Depends on what sense of "shared caches" - I can read it two ways:
Sense 1 : everything shares a single large resource for caching - so
caching concentrates on the heaviest user. This is something I want to
try sometime but other things are ahead of it in the dev queue. Feel
free to experiment - it's easiest with direct mode (not sure it makes
much sense for mapped mode)
Sense 2: They share the data in the cache. Maybe for the node table but
they have to be built together; unlikely to help for the indexes.
Andy
