---
title: "Discussion on \"DuckDB 2.0 on a 2 vCPU box: recursive queries 20x faster, VARIANT 80x\" | Hashnode"
description: "Discussion on \"DuckDB 2.0 on a 2 vCPU box: recursive queries 20x faster, VARIANT 80x\". I reran DuckDB 2.0 benchmarks on a 2 vCPU box: a recursive ancestry walk went from 426 to 21 ms, VARIANT queries from 1,315 to 16 ms."
url: "https://hashnode.com/posts/duckdb-2-0-on-a-2-vcpu-box-recursive-queries-20x-faster-variant-80x/6accddabf3c3aed6bd0fd5fd"
---

[Feed](https://hashnode.com/feed) /

Discussion

# [DuckDB 2.0 on a 2 vCPU box: recursive queries 20x faster, VARIANT 80x](https://axionry.hashnode.dev/duckdb-2-0-on-a-2-vcpu-box-recursive-queries-20x-faster-variant-80x?utm_source=hashnode&utm_medium=feed)

I reran DuckDB 2.0 benchmarks on a 2 vCPU box: a recursive ancestry walk went from 426 to 21 ms, VARIANT queries from 1,315 to 16 ms.

Oct 12 · 5 min read · axionry.hashnode.dev  
Read the article

-   [#duckdb](https://hashnode.com/tag/duckdb)
-   [#sql](https://hashnode.com/tag/sql)
-   [#data-engineering](https://hashnode.com/tag/data-engineering)
-   [#python](https://hashnode.com/tag/python)

## Written by

[Neeraj Sharma](https://hashnode.com/@neeraj0786)

@neeraj0786 · Gurugram, India

  
[View profile](https://hashnode.com/@neeraj0786)

Founder @ axionry.com | Head of Engineering @ AccioJob (YC'19)

I'm Neeraj Sharma, founder of Axionry (https://axionry.com), an AI product studio building AI products, AI agents, voice agents and MVPs that ship in days, with $0 upfront. Head of Engineering at a YC-backed company: I run 4 to 5 large platforms with a small team, cut a $200K/year cloud bill by more than 70% and took voice AI from 10 cents to 2.5 cents a minute. Book a call: https://axionry.com/book | X and Instagram: @axionry\_

## Responses

Markdown supported · ⌘↵ to post

No responses yet.