Apache Arrow files

Apache Arrow is the columnar memory format that most modern data tools share. AddisDB reads Arrow IPC files directly, keeping the column types exactly as they were written.

Who it is for

Arrow is the reason a dataframe can move between Python, R, Rust, Java and a database engine without being converted at every boundary. Its IPC files are that same layout written to disk.

The formats around it are worth telling apart: Arrow is the in-memory layout, Feather v2 is that layout on disk, and Parquet is a compressed storage format that has to be decoded into Arrow to be used. Arrow is the fastest to read and the largest on disk.

Use it when your pipeline already speaks Arrow, when you want zero-cost handoff between tools, or when you have an .arrow file from a system and want to look inside it.

Producing a file

# pyarrow
import pyarrow as pa, pyarrow.ipc as ipc
with ipc.new_file("data.arrow", table.schema) as w:
    w.write_table(table)

Open it in AddisDB

1. New Connection → Files → Feather → Apache Arrow.
2. Choose the .arrow file, name the connection, and Save.

What AddisDB gives you

• SQL over the file with types preserved from the Arrow schema.
• The Chart view, notebooks, and AI chat grounded in the real columns.
• ⌘K search across every connection.
• A file that is only ever read — nothing is written back to it.