I have tried uploading the simple logs data of windows from direct upload with about 279 KB file size that after ingestion into elastic was about 277.58kb.
named windows_2k_simple that 277.58kb.
same with logsdb setting through index.mode:"logsdb"
The Hierarchy of an index is
Index -> Shards -> Segments
When you first start writing data to elasticsearch it automatically creates a number of segments to "get ready" to ingest data... at that point there is more overhead than actually data.
You will see the logsdb savings when you scale up ... Event at 1GB or more but definitely at 10Gb.... etc
To be clear there is not a "guaranteed" saving %.... anywhere from 30-70% depending on the type / content of logs... I see mostly 40-60% savings with the data sets I have tested
Also adding that you need also a trial or enterprise license to get the maximum compression ratio. Otherwise you will see some benefits between without logsdb and with it, but not like 40-60% because synthetic source is only available with a trial/enterprise license.
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.