Load Balancer, fail tolerance and replication in PostgreSQL

["Miguel González Castaños" <mgc@tid.es>]

Dear all,

  We have a web application served by Apache with Tomcat behind for the 
JSPs. The database backend is PostgreSQL. I have been asked to do some
research about how to achieve fail tolerance and load balancing in our 
application. After searching a little bit this is my first impression:

  - There are integrated software for high availability using PostgreSQL 
(i.e.: eRserver) but have commercial licences (you have to pay a lot for 
it).

  - There are some efforts for developing GPL application to achieve 
replication, or load balancing or even clustering. My impression is that 
they are in a earlier stage, but some of them look promising. Anyway I 
am not very sure if any of them will suit into my setup (like pgcluster 
or dbmirror).

  What I have thought that in my setup maybe I just need to do 
clustering  with servers with the same setup: Apache/Tomcat. Then, just 
do some sort of replication between the PostgreSQL databases that every 
node of the cluster should have (i dont know which is the best GPL 
replication software). Even I thought that maybe I could use some sort 
of dns round- robin approach for load balancing for Apache.

  Anyway as you can see I´m pretty lost and I need some guidance to put 
me in the right direction...

  Sorry for my English

  Many thanks in advance

  Miguel







Se que en Bulma sois mas forofos de MySQL que de PostgreSQL por
vuestras experiencias con la web de bulma, pero tengo que hacer un
estudio para usar PostgreSQL en un entorno donde se necesita balanceo de
carga y tolerancia a fallos.

Realmente lo que tenemos es una aplicacion web, con Apache y Tomcat
como servidores de JSPs cuya base de datos es PostgreSQL, digo esto
porque he pensado que tal vez lo de buscar un sistema de alta
disponibilidad para PostgreSQL no sea necesario para este sistema, y
simplemente lo que podria hacerse seria un cluster con Apache, o incluso
utilizar un algoritmo round-robin que vaya pasando las solicitudes a los
distintos servidores y un sistema de mirroring entre los distintos
servidores PostgreSQL.

He estado mirando en la página web de PostgreSQL y navegando un poco, y
veo que hay alternativas que ofrecen alta disponibilidad (tolerancia a
fallos, balanceo y replicacion), pero de pago.

Como os digo se me ocurre a priori este enfoque que os digo, ya que
realmente solo hay una base de datos que mantener sincronizada (creo que
esto se puede hacer con dbmirror aunque no se si es solo a nivel de
tablas o de bases de datos).

Tambien he visto una herramienta llamada pgcluster. La verdad estoy un
poco perdido, alguien me puede echar un cable?

Un saludo

Miguel



-
To unsubscribe from this list: send the line "unsubscribe linux-admin" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html