The replication was never as stable as required and it’s very hard to extend it on additional properties or requirements. Furthermore the search in AEM is quite difficult due to the complex dependencies.
Against this, the usage of Solr3 as intended by the Jackrabbit Oak4 developers has many advantages. First of all the configuration is a no-brainer, followed by the out-of-the box indexing functionality. Given the provided schema and configuration5 and examples, you should be able to get a search page up and running in a very short time.
All further required configurations regarding indexing behavior can be done either in CRX or in the Solr server and don’t require any code-changes.