I was looking into RAM to Disk Ratio and for logging uses cases and I see the proportion as 1:48 to 1:96. So if the server has 128GB RAM can we consider 96*128GB RAM = 12 TB Of disk space as a max size per node?
Often the amount of data you can hold on a node depends on how efficiently you can manage heap usage, which is a finite resource. As we recommend to have a heap of ~31GB or 50% of RAM (whichever is smaller), having 128GB RAM per node does not necessarily allow you to store more data on the node than if it had 64GB of RAM, although there will be more memory for the file system cache which may make queries faster.
The ratios you have quoted are guidelines that work for a reasonably wide range of use-cases, and usually assume up to 64GB RAM per node. This is however not an absolute limit and depending on your use case your maximum ratio can differ.
For ECE the maximum node size is generally around 64GB and you instead up having multiple nodes per 128GB host.
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.