As part of our on going engagement with Tesco.com, we were tasked with optimising the Endeca indexing process of the UK Home Grocery Shopping (GHS) Endeca deployment. The overall Product Induction is a multi-step multi-hour long process that runs on a daily basis to update the UK GHS (http://www.tesco.com/groceries) website with the latest product set, pricing, stock availability, promotions etc. The last step in this process is the Endeca indexing process, that refreshes the Endeca search & navigation engines that power the UK GHS website.
In order to find areas of potential optimisations, Excogis analysed the Endeca deployment as a whole, including undocumented legacy code from the original deployment 7 years ago. Following this analysis, our team re-structured the Endeca Information Transformation Layer (ITL) configuration that controls the processing of the source data into Endeca data records, that are in turn indexed into a binary structure used by the active search & navigation engines. This re-structuring involved the following areas of improvement:
- Converted old embedded Perl script code that was resource intensive into more efficient custom Java components.
- Streamlined the manipulations of the source data to avoid multiple reads/processing of the same records.
- Created a new Java-based multi-threaded custom adapter to read source data records in parallel.
- Eliminated unnecessary data manipulations.
The result of these optimisations was the reduction of the Endeca data processing time by approx. 45%. This allows Tesco to make a refreshed website available to users even earlier in the day, benefiting both the UK GHS customers and the Tesco.com IT Operations team.
To learn how we can help you improve your own Endeca deployment, contact us at firstname.lastname@example.org.