Years ago, we started out using Solr as a Data Indexer and only later used Elasticsearch as a Datamart. More recently — as we have been scaling Germain to support our ever-growing client’s needs — we faced a number of challenges:
Scalability limitations with Solr.
Many advanced features we rely on are not, or only poorly, supported by Solr. We had to limit our capabilities to the lowest common denominator between the two products.
Had to maintain two separate versions of our indexer code for Solr and ES; this increased code complexity, maintenance and QA effort.
Now, all this is history. We no longer support Solr, and rely on ElasticSearch instead as a Datamart and Data Indexer.
This switch to ES reduces the overall number of services needed by Germain.
System requirements remain the same; any CPU / RAM previously set aside for Solr can be used for ES instead.
For customers currently using a relational database (MySQL, Oracle, etc.) for their Datamart, they can continue to do so. For the next upgrade, we will switch out Solr for ES.