Term complexity and filter caching

Hi peeps,

Leaf level filter cache is nice, but caching complex filters is better if
you can get some hit ratio.

We have some items (querystrings that may become bool filters later) with
over 50 terms (sometimes hundreds) that get reused often. We essentially
aggregate financial research for hundreds of sites, each of which have only
paid for certain data and this is the primary case for these complex and
highly reused queries.

However, we are looking at adding in some leaf level filter caching on
other items. Say an industry or a report type. Across fields with small
permutation and high re-use.

Elasticsearch employs an LRU policy on filter cache.

However, the complex queries with high term complexity and lots of "AND"s
and "OR"s saves far more time, CPU and memory per use than the leaf level
caching on something like "reporttype:research" , so I'd like to add some
weight to the more complex queries when determining what gets evicted.


--Shannon Monasco

You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/e51fafea-d4f3-4bc3-823e-26f9e40dfe89%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.