On 7/18/22 02:55, Daulat wrote:
Hello Team,
We are planning to create multiple clusters on a single server (single
instance) with PostgreSQL V.10 to run multiple applications.
Version (9,6, 10, 111, 12...) does not matter.
I don't know if it is a good idea to use a single machine to run n
clusters but we are looking to minimize the cost of servers and other
resources.
That depends on your circumstances.
Please share your thoughts on this approach?
It allows you to do PITR backups without doing all-or-nothing restores.
How can we create multiple clusters under a single postgres version?
Separate $PGDATA directories, and separate postgresql.conf files (with each
config file specifying a different port number: 5432, 5433, 5434, etc)
How can we identify/allocate the resources to parameters like
shared_buffers, effective_cache_size, work_mem etc in case of multiple
clusters?
Very trickily.
--
Angular momentum makes the world go 'round.
