The application that I am currently working on does not need real time indexing. I am using Solr as the indexing engine and I want to setup a High Availability Solr cluster with 2 replica nodes. Seems zookeeper can be of help, but again, to make it HA i need to spend some infrastructure for zookeeper replication itself. The budget constraint makes me think of an alternative solution.
As my application does not need real time indexing, I am thinking of developing a daemon which can keep a queue of indexing data and post it to both the instances of solr. To provide load balancing & failover, I will use the same HAProxy load balancer that I am using for my current Application servers.
Does this configuration makes sense?