Skip to main content

Iara

A parallel query engine for analytics on a single machine. The core is written in Rust on Apache Arrow. It is used from Python and Node.js, through a lazy DataFrame API and SQL, and from Iara Studio, a web workspace for notebooks and SQL.

import iara
from iara import functions as F

revenue = (
    iara.scan_parquet("data/lineitem")
    .join(iara.scan_parquet("data/orders"), left_on="l_orderkey", right_on="o_orderkey")
    .group_by("o_orderpriority")
    .agg(revenue=(F.col("l_extendedprice") * (1 - F.col("l_discount"))).sum())
    .sort("revenue", descending=True)
    .collect()
)

iara.sql("select k, sum(v) from t group by k", t=arrow_table)   # pyarrow, pandas or polars inputs

What it does

  • Data sources:
    • Parquet, CSV, JSON and Delta Lake tables;
    • local files or S3-compatible object stores (S3, R2, MinIO);
    • Hive-partitioned datasets.
  • SQL and Python:
    • SQL in a DuckDB-like dialect, and a Polars-like DataFrame API;
    • joins, window functions, recursive CTEs;
    • lists and structs, schemas;
    • SQL parameters (:name, $1, ?), views, TABLESAMPLE, DISTINCT ON, SELECT * REPLACE/RENAME/EXCLUDE;
    • writes: COPY, CREATE TABLE AS, INSERT, and on Delta tables UPDATE, DELETE, MERGE, TRUNCATE.
  • Execution:
    • uses every core;
    • spills to disk when a query outgrows its memory limit;
    • streams results without collecting them.
  • Studio: reactive notebooks stored as plain .py files (marimo format), SQL files, a data catalog, query profiles, charts, dashboards, scheduled jobs and on-demand clusters. Sign-in and per-user data grants are optional.

Install

With uv:

uv add iaradb            # in a uv project
uv pip install iaradb    # in any virtual environment

pip install iaradb works too. The package is named iaradb on PyPI and imported as iara.

Wheels are published for Windows and Linux (x86-64), Python 3.10 and newer. On other platforms, build from source: see docs/development.md.

Performance

Sum of the query times, on the same Parquet files (AWS m7i.4xlarge, 16 vCPUs, 64 GB RAM, measured 2026-10-22). Iara is compared with DuckDB and Polars here; seven engines in total are in the docs:

Suite Iara 0.1.0 DuckDB 1.5.6 Polars 2.0.0
TPC-H SF10, all 22 queries 3.46 s 5.37 s 6.09 s
TPC-H SF100, all 22 queries 36.6 s 58.5 s 69.3 s
ClickBench, all 43 queries 17.8 s 26.0 s 28.9 s

TPC-H results were checked against DuckDB's. Per-query numbers and the other engines: workhouse.dev/benchmarks.

Documentation

License

Apache-2.0

Metadata

Release files for iaradb 0.1.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Built distributions (wheels)

Table of built distributions (wheels) for iaradb 0.1.0
File Interpreter ABI Platform Actions
iaradb-0.1.0-cp310-abi3-win_amd64.whl CPython 3.10 abi3 Windows x86-64 Details for iaradb-0.1.0-cp310-abi3-win_amd64.whl
iaradb-0.1.0-cp310-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl CPython 3.10 abi3 Linux glibc 2.17+ x86-64 Details for iaradb-0.1.0-cp310-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl

Total release size: 34.2 MB

Release files / iaradb-0.1.0-cp310-abi3-win_amd64.whl

Download URL iaradb-0.1.0-cp310-abi3-win_amd64.whl
Size 16.7 MB
Tags CPython 3.10 Windows x86-64 abi3
SHA-256 checksum
How to use checksums
c90f50b69b6854ef2a67ba1e46f2ad5ac9cf60fe18388ffb3201631a4a8bca14
BLAKE2b-256 checksum
How to use checksums
96242115859e8d162d949e8f94868691409869e5fa58359593f8d13254b921e5
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via uv/0.9.5

Release files / iaradb-0.1.0-cp310-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl

Download URL iaradb-0.1.0-cp310-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
Size 17.5 MB
Tags CPython 3.10 Linux glibc 2.17+ x86-64 abi3
SHA-256 checksum
How to use checksums
f4c3781f1210f70a443e7918d0f59adc390a068037cd015896e6378950cc542c
BLAKE2b-256 checksum
How to use checksums
c9992408e9d2402b063f250429a314ccc35a81998dd209f8ca1f4b02fb2d7911
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via uv/0.9.5

Security for iaradb 0.1.0

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Release history Release notifications | RSS feed

This release

0.1.0 This release

2 release files No provenance

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page