Watching this recent talk frozen indices sounds like a feature recommended to users
However when I look at the doc here the feature is deprecated.
I'm not sure which one to read into - Is the feature going to exist going forward?
If it's been deprecated I'd be curious to know the cost angle involved here - With frozen indices we could use S3 which is 4x cheaper than EBS if we did not have frozen indices support - for data that is infrequently accessed what would the recommended approach be to replace frozen indices with
First I want to apologize for the confusion ... the Term "Frozen" has 2 meanings, the old meaning is going away.
1st The "old/deprecated" Frozen Indices which is now deprecated, was basically a method to minimize the heap space used by indices that were rarely accessed but still resided on local disk. Put that one out of your mind.
Somewhat unfortunately we used the same term to apply to the New Frozen Tier (I think it is what we and our customers really wanted all along)
The "new" Frozen Tier is what you want to focus on which uses the Capability of Searchable Snapshot to allow searching vast amount of Elasticsearch data that resides as snapshots on cheap Objects Stores such as S3. This is the one you want to pay attention to.
That is the one on the webinar that you watched.
Here is another article on it:
Hope that helps... You're not the first one (and probably not the last) to be a bit unclear on this.
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.