ELMo-Tune-V2: LLM-Assisted Full-Cycle Auto-Tuning to Optimize LSM-Based Key-Value Stores
Viraj Thakkar 1 , Qi Lin 1 , Kenanya Keandra Adriel Prasetyo 2 , Raden Haryosatyo Wisjnunandono 3 , Achmad Imam Kistijantoro 3 , Reza Fuad Rachmadi 2 , Zhichao Cao 1 1 Arizona State University, 2 Institut Teknologi Bandung, 3 Sepuluh Nopember Institute of Technology and · 2025
Read ELMo-Tune-V2: LLM-Assisted Full-Cycle Auto-Tuning to Optimize LSM-Based Key-Value Stores by Viraj Thakkar 1 , Qi Lin 1 , Kenanya Keandra Adriel Prasetyo 2 , Raden Haryosatyo Wisjnunandono 3 , Achmad Imam Kistijantoro 3 , Reza Fuad Rachmadi 2 , Zhichao Cao 1 1 Arizona State University, 2 Institut Teknologi Bandung, 3 Sepuluh Nopember Institute of Technology and (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2502-17606.txt.
Description
Viraj Thakkar, Qi Lin, Kenanya Keandra Adriel Prasetyo, Raden Haryosatyo Wisjnunandono, Achmad Imam Kistijantoro, Reza Fuad Rachmadi, and Zhichao Cao let a large language model tune database engines by itself. Their 2025 paper concerns LSM-based key-value stores, the storage backbone of many modern systems. These engines expose a huge configuration space covering compaction, flushing, and caches, all competing for the same CPU, memory, and storage.
Human tuning is slow and expert-bound; earlier auto-tuners assumed one workload and touched few knobs. Their framework, ELMo-Tune-V2, uses the model's reasoning to characterize workloads on its own, tune broadly across parameters, and adjust settings live as conditions change. The paper argues language models can carry real systems work, not just text tasks.
Catalog record
| CREATOR | Viraj Thakkar 1 , Qi Lin 1 , Kenanya Keandra Adriel Prasetyo 2 , Raden Haryosatyo Wisjnunandono 3 , Achmad Imam Kistijantoro 3 , Reza Fuad Rachmadi 2 , Zhichao Cao 1 1 Arizona State University, 2 Institut Teknologi Bandung, 3 Sepuluh Nopember Institute of Technology and |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 60 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-17606