Thank you David and Amit. This is more or less what I was looking for.
I _think_ I might be able to store the data as TEXT, which is highly compressed by Toast, and then perhaps write the function in terms of a TEXT to JSONB conversion. I will give it a try. It might perform terribly, but will be an interesting experiment.:)
> Hi, > > I would like to create a GIN index on a set of JSON documents. Right now > I'm storing the data in a JSONB column. The current index looks like this: > > CREATE INDEX document_payload_idx > ON document > USING gin > (payload jsonb_path_ops); >
> The index is pretty small, but the actual data takes up a *lot* of space. > Is there a way to get Postgres to index the table *as if* the JSON were
> there, but not actually put the data in the table? I could either store > the > docs elsewhere and keep a reference, or compress them and put them in the > table in compressed form as a blob. > > Thanks much for your help. > > Larry
No idea if this works but maybe you can store the compressed data and then write the index expression like:
USING gin (unzip(payload) jsonb_path_ops)
The unzip function would need to be custom I think...
