Don't have much experience with Presto, but I have used Hive to query Elastic Search. It works very very well for full-table-scan analytics-type queries. I exp

Don't have much experience with Presto, but I have used Hive to query Elastic Search. It works very very well for full-table-scan analytics-type queries. I expect PrestoDB would be similar. But when it comes to queries about smaller and smaller pieces of the full dataset, it becomes less and less likely that these types of connectors perform well. Predicate pushdown is rarely well-implemented in these types of "run SQL against any big data" systems (Hive, Presto, Impala, SparkSQL, etc). A simple "select * where id = 1234" will often do a full scan and filter within the query engine, rather than push the point lookup into ES.