It depends on your data volume and growth how you can balance between cost and performance. For ongoing operations, you should be able to provide an I/O subsystem with very low latency and high IOPS. Elasticsearch is very sensitive to this. For my relatively small data set of 1 TB, VM or SAN is always second choice. I prefer the physical machine with custom file system setup on tier 0 or tier 1 storage (preferably SSD) for maximum performance.
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.