I am building a web app backed by database postgresql and currently the
site is under a humongous torrent of hits. I have a question regarding
the database backend config.
The database folder has a file called postgresql.cons where we can
specify no of active connections to the database -N which is currently
1024 and also specify the shared buffers. Is there any thumbrule in
deciding the -N and also the no of shared buffers for my config of 4
frontend servers and 1 database backend server. There is no global or
memcached mechanisms in use.
I am building a web app backed by database postgresql and currently the
site is under a humongous torrent of hits. I have a question regarding
the database backend config.
The database folder has a file called postgresql.cons where we can
specify no of active connections to the database -N which is currently
1024 and also specify the shared buffers. Is there any thumbrule in
deciding the -N and also the no of shared buffers for my config of 4
frontend servers and 1 database backend server. There is no global or
memcached mechanisms in use.
Any pointers would be greatly appreciated.
You're on the wrong list for this... head over to the postgresql mailing lists if you haven't already... it's not so much a matter of how many front end servers you have as what the queries are, dataset size, ordering, etc, etc, etc...
I am building a web app backed by database postgresql and currently the
site is under a humongous torrent of hits. I have a question regarding
the database backend config.
The database folder has a file called postgresql.cons where we can
specify no of active connections to the database -N which is currently
1024 and also specify the shared buffers. Is there any thumbrule in
deciding the -N and also the no of shared buffers for my config of 4
frontend servers and 1 database backend server. There is no global or
memcached mechanisms in use.
Any pointers would be greatly appreciated.
AFAIK the norm for rails apps is one db connection per mongrel/fastcgi instance.