ParqDB
1 commentThe first launch from ParqDB
ParqDB publishes a vector index to object storage as immutable Parquet files and then searches it from the browser, reading only the byte ranges it needs. There is no query server in the middle. The public demo searches a hundred thousand Wikipedia articles this way.
Meet the team
What people who use it say
Nobody has reviewed ParqDB yet. If you have used it, you are the one who can say whether it holds up.
Write the first review
Sign in to comment
Search with no server to run is a category of idea that usually collapses on contact with a real index size. This one holds up on a hundred thousand documents, in a tab, over range requests.
Whether it stays sane at ten million is the question, and immutable Parquet on object storage is a promising place to find out. For a static site with a corpus that changes weekly, it might already be enough.