In March 2015, Shay Banon announced the deprecation of rivers in Elasticsearch due to their contribution to cluster instability, which stemmed from their reliance on external systems and libraries. While initially successful in facilitating data integration into Elasticsearch, rivers introduced overheads and bugs that compromised system resiliency. The decision to phase out rivers, starting from version 1.5, was made to encourage more stable methods of data ingestion, such as using official client libraries or tools like Logstash, which can also munge data before it enters Elasticsearch. Despite the success and community involvement in developing river plugins, the focus has shifted towards using external processes or integrating with Logstash inputs, offering a smoother transition for developers and maintaining ease of data ingestion. Banon acknowledged the community's efforts and expressed a commitment to assist developers in adapting their river plugins through alternative means.