The optimum performance of any business-critical application demands excessive attention from techies and time-to-time optimization also play significant role keeping the application’s performance at its best. Recently, I explored a possibility of optimizing a full functioning Liferay application simply through removing a technology tool Google Search Appliances (GSA) from it. This blog details out the primary objectives and impacts for the same.
The Liferay application provides domain specific scientific data to its users based on the predefined search criteria and parameters. Performance optimization was required for aspects such as content uploading time and page loading time. The application accommodates massive amount of contents in various forms searchable through excessive flexible search function. Both Google Search Appliances (GSA) and Apache Solr were integrated as the search engines for indexing as well as front end displaying of the data.
Though GSA can certainly be considered as one of the best available tools to furnish the result relevancy, we decided to remove it so that we can explore:
- If Apache Solr can handle the desired definitions of search functions or not
- Using single search engine can improve the performance of the application or not
While removing integration of GSA, we paid extra attention to the following criteria:
- Scalability of the application
- Content processing
- Data indexing
- Query functionalities
- Search relevancy
- Administration & maintenance
Advantages of removing GSA
 Saving of time in uploading contents: As one entire layer of indexing is removed from application, adequate amount of time can be saved while uploading the contents. Earlier, the in general time for uploading contents was about 8 to 30 minutes. It decreased up to 2 to 5 minutes, even when the content get pushed from staging environment to PROD environment
 Performance improvement: Overall page loading time decreased up to 20-50 %
 Cost Saving: GSA is a paid tool and for this application, its annual cost was around $50,000
I believe removing the integration of GSA turned out to be an enriching experience. We also developed certain specific features to match with precise requirement of the application’s search function attainable through Apache Solr. We could notice statistically verified improvement in almost each aspect of the application’s performance parameter providing an undisrupted search experience to internal as well as external users.