Open
Description
At the moment Indexify stores data in S3. Most of the times downstream functions will immediately read them, we could store the outputs on SSD of the executor which ran the function and serve the output to the edges directly without a round trip to S3.
This could be useful in Document AI. We spend about 4-8 seconds on every functions doing I/O from S3 at the moment.
Metadata
Metadata
Assignees
Labels
No labels