Here are the reasons
1. Eleasticsearch needs to maintain document in int own database Whereas sphinxsearch reads the data builds the index and discards the document. This means I dont have to replicate the data from anahita database to Elasticsearch database
2. In current architecture for Anahita, I found out that I can use all the functionality and facets without making much modification as long as I can feed node IDs to search code. I could be wrong but sphinxsearh integration appears more seamless than elasticsearch
3. Popularity of search engine was one of the criteria however it was not the criteria for me while selecting search engine. My primary criteria was unicode search, soundex and prefix/suffix functionality. I could locate equal amount of information on web for eleasticsearch as well as sphinxsearch. From functionality perspective, I couldnot find any thing specific in elasticsearch which cannot be sayisfied with sphinxsearch.
4. Last but not the least, I was trying to avoid any java based searchengine as memory requirements are typically high in Java. I was looking for a search engine which is fast enough but with smaller footprint. Now I could fit mysql, php with APC, anahita code, varnish and sphinxsearch all in 2GB machine on rackspace and still supporting approx. 30000 daily visitors with a peak of 250 loged in users and 6 concurrent transtions.

Please note that selection of sphinxsearch was based on above mentioned specific needs and it may not be applicable to your project(s) Read Less

Ali, Couple of months back I wrote an article on sphinxsearch -

I'm planning to write another one about integrating it with Anahita . Hopefully I can finish it this weekend.

Summary is - Modify the where clause in anahita/src/site/components/com_search/domains/queries/node.php and call sphinxsearch PHP call to get ID

You can find sphinxsearch PHP API details at - http://sphinxsearch.com/wiki/doku.php?id=php_api_docs Read Less