Corning

Option A – Pure Fusion Search

Overview

The main feature of this architecture is that all searches would originate from a single collection and there’s only a single SolrCloud serving production queries.

Benefits

Only have to manage a single SolrCloud and it can be the latest version supported by Fusion
Only have to manage a single collection for all 3 types of search, regardless of which Ul the search is coming from (exception: side car collections L.e. for signals).
Hybris can use it’s own, bootstrapped Soir instance (doesn’t matter if it is cloud or not). All that matters is that Fusion can see it in order to copy the documents into the Unified Search
collection. Hybris embedded Solr would serve as a “staging” index.
Doesn’t require building installing any plugins into Solr
Doesn’t require any modifications or hacks to Hybris’ indexing.

Trade Offs

Sull have to maintain a Solr instance for Hybris to do its indexing, but it can be smaller and doesn’ require HA, since it will only be used for indexing and will not serve any end-user queries.
Requires modifications to the Hybris Ul to use a Fusion Query Pipeline endpoint instead of the usual Solr end point. This is already assumed.