Обсуждение: pgsql cluster
Hello, My name is Zsolt and I would like to ask for some advice, because I'm novice in this topic. We would like to install our webpage based on postgresql to our server. We have the following architecture : 2 load balancer, 2 ibm server, 1 shared storage. The shared storage has redundant controllers and the storage is connected to the servers through redundant optical cable. The loadbalancer is redundant too. The load balancers choose the less loaded server to serve the webpage requests. I have gone throught the postgresql documentation as I see there is two clustering solution. - pgpool (http://pgpool.projects.postgresql.org/) - pgcluster (http://pgcluster.projects.postgresql.org/1_3/pglb_conf.html) What I need : - one ip address/port to connect from webapplication - two postgresql server running on the 2 ibm server accesing the same database stored on the shared storage. Which clustering solution is the better for me? Can I setup two loadbalancer to be redundant? Can access the same database on the shared storage the two postgresql server? Thank you for your help! Best Regards Zsolt
On Friday 20 November 2009, Démi Zsolt <zsdemi@gmail.com> wrote: > Which clustering solution is the better for me? Can I setup two > loadbalancer to be redundant? Can access the same database on the > shared storage the two postgresql server? Well, if it's an important application, you should probably hire someone who knows what they're doing to set it up; this stuff is complex and easy to get wrong. But the short answers are: Yes you can setup 2 load balancers to be redundant, there are several clustering solutions available for Linux or other operating systems. Linux-HA is a good place to start. No, you cannot run 2 instances of PostgreSQL against the same physical database. You could use Linux-HA tools (ie. heartbeat) to make an active/passive cluster of the 2 database/web servers, though. Proper fencing is going to be important on shared storage. Make sure you're also doing real backups of the database.