Running Elasticsearch completely in memory


I am trying to find if I can run Elasticsearch completely from physical RAM like a in-memory DB. My index size is relatively small (lets says ~10GB) and is more or less constant size and does not change that frequently and data comes from a SQL Server, is it possible to run this Elasticsearch cluster completely from memory? I understand the risk that a node failure will need reindexing, if I dont have replicas. If I am okay with that constraint, is there a way to make this run completely in-memory.

We will be using Elasticsearch v5.x on Windows 2008 R2 server.


You can, but via a memory disk that you need to manage via the OS.

ES is no longer able to do this.

Thanks for the response. Is there any reference documentation on how to do this? What kind of configuration needs to be setup.

We don't have anything on the OS level stuff, but once it's done just point ES to it as you would any other non-default mount, ie

If your data set is small enough to fit in the file system cache, you should be running entirely in memory even with default setting as all data files will be cached in memory.

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.