Re:Building a streaming SQL engine with Arrow and DataFusion
by jnewhouse on Monday March 18, 2024 @01:25PM
(#39747225)
Yep, pretty much. Right now filesystem^ sources are finite, scanning the target path at operator startup time and processing all matching files. This processing is done by opening an asynchronous reader, courtesy of the object_store crate.
^We call these Filesystem Sources/Sinks to match terminology present in other streaming systems, but I'm not in love with it.
So if I'm understanding, you actually read data directly from (say) S3? It isn't copied from S3 and stored locally (ie, a bunch of local .arrow files.)
(Apologies if I'm ignorant of the underlying tech - I think this is really cool and just trying to wrap my head around what happens from "I upload some data to S3" and "we get query results")