writing indices in remote filesystem

[kyogesha]

Hi,
Is it possible to write indices in remote filesystem like s3 (possibly in future)?

[sezruby]

Though I haven't tested yet with a s3 path, you can change the path for index data using

spark.conf.set("spark.hyperspace.system.path", "/path/to/indexes")

Then your index (indexName) will be written in /path/to/indexes/indexName at index creation.
Note that Hyperspace tries to read indexes from the path when query time, so it's not possible to use multiple index data paths for now.