pith. sign in

arxiv: 2605.19988 · v1 · pith:ITA2SKYInew · submitted 2026-05-19 · 💻 cs.SE · cs.AI· cs.DB· cs.PF

A Case for Agentic Tuning: From Documentation to Action in PostgreSQL

Hongyu Lin , Mingyu Li , Weichen Zhang , Yihang Lou , Mingjie Xing , Yanjun Wu , Haibo Chen This is my paper

Pith reviewed 2026-05-20 03:36 UTC · model grok-4.3

classification 💻 cs.SE cs.AIcs.DBcs.PF
keywords PostgreSQL tuningLLM agentsperformance optimizationsystem configurationdatabase benchmarksagentic tuning
0
0 comments X

The pith

Translating expert tuning knowledge into executable skills lets LLM agents optimize PostgreSQL more effectively than static documentation.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

Documentation for system tuning records only the final recommendations from experts while discarding the reasoning steps that produced them. This creates guides that quickly become outdated with new software versions, perform poorly on different workloads, and miss how multiple parameters influence one another. The paper presents PerfEvolve as a way to convert those expert methodologies into executable skills that equip LLM-based agents to check version consistency, profile specific workloads, and jointly optimize several parameters at once. On PostgreSQL with standard TPC-C and TPC-H benchmarks, this approach yields measurable gains over existing documentation-based methods.

Core claim

PerfEvolve translates expert tuning methodologies into executable skills that enable LLM-based agents to perform version-consistency verification, workload-specific profiling, and multi-parameter joint optimization. Evaluated on PostgreSQL under TPC-C and TPC-H benchmarks, PerfEvolve outperforms state-of-the-art documentation-driven tuning baselines by up to 35.2%.

What carries the argument

PerfEvolve, the system that converts expert tuning methodologies into executable skills for LLM agents to act dynamically on system configuration.

If this is right

  • Version-consistency checks become an automated agent step rather than a manual review.
  • Profiling can adapt directly to the current workload instead of relying on generic advice.
  • Joint optimization across parameters accounts for dependencies that static guides overlook.
  • Tuning remains usable as the software evolves without requiring constant manual updates to documentation.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same skill-translation process could be applied to tuning other databases or operating systems.
  • Agents might eventually generate new skills from observed performance data rather than only from existing expert methods.
  • This raises the practical question of how to audit or correct skills when an agent makes an unexpected configuration choice.

Load-bearing premise

Expert tuning methodologies can be reliably turned into executable skills that agents follow without missing key parameter interactions or using outdated information.

What would settle it

Deploy the agent on a newer PostgreSQL version with a workload that has strong inter-parameter effects and check whether it still selects the same parameter settings as the documentation baseline or produces lower performance.

Figures

Figures reproduced from arXiv: 2605.19988 by Haibo Chen, Hongyu Lin, Mingjie Xing, Mingyu Li, Weichen Zhang, Yanjun Wu, Yihang Lou.

Figure 1
Figure 1. Figure 1: Latency increase on TPC-H when applying PG-Official and PGTune rules (7 of 22 queries [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Independent vs. joint optimization for shared_buffers (sb) × checkpoint_completion_target (cct) across three workloads. “Independent” tunes each parameter to its individually-optimal value in isolation, then combines the individually optimal values. “Joint” searches the two-dimensional space together. Finding-#2: Optimal configurations are workload-dependent. System documentation typically provides generic… view at source ↗
Figure 3
Figure 3. Figure 3: Interaction strength for 55 parameter pairs on TPC-C-r. Each cell shows the interaction [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: PERFEVOLVE architecture. The offline profiling phase runs once per system version: it applies dimensionality reduction and topology discovery to produce a procedural tuning document. The online deployment phase runs per deployment: an LLM agent executes the document’s skills, performing targeted profiling guided by the offline-derived reference data. Goals. PERFEVOLVE is designed around three objectives: •… view at source ↗
Figure 5
Figure 5. Figure 5: Sensitivity (CV) of all 116 PostgreSQL performance-relevant parameters sorted in descend￾ing order, on the TPC-C-r workload. The distribution exhibits a pronounced long tail: the top 15 parameters account for >95% of cumulative performance variation. The dashed line marks the 5% threshold τs. 3.2 Correlation Topology Discovery Problem statement. Reducing to k parameters is necessary but not sufficient: if … view at source ↗
Figure 6
Figure 6. Figure 6: Correlation graph for PostgreSQL on the TPC-C-r workload. Left: nodes are the top￾15 parameters; edges indicate confirmed correlations (η 2 > 0.15, p < 0.05); edge thickness is proportional to η 2 ; colors denote connected components. Right: the resulting optimization strategy. Component C1 (shared_buffers, work_mem, checkpoint_completion_target) requires a 3- dimensional joint search; C2 requires 2-dimens… view at source ↗
Figure 7
Figure 7. Figure 7: Profiling cost vs. tuning quality. Each stage completion triggers a discrete jump in [PITH_FULL_IMAGE:figures/full_fig_p015_7.png] view at source ↗
read the original abstract

Documentation has long guided computer system tuning by distilling expert knowledge into per-parameter recommendations. Yet such guides capture only what experts conclude, discarding how they reason. This fundamental gap manifests in three concrete deficiencies: documentation grows stale as software evolves, fails under heterogeneous workloads, and ignores inter-parameter dependencies. We propose shifting from static documentation to dynamic action for system tuning. We introduce PerfEvolve, which translates expert tuning methodologies into executable skills that equip LLM-based agents to perform version-consistency verification, workload-specific profiling, and multi-parameter joint optimization. Evaluated on PostgreSQL under TPC-C and TPC-H benchmarks, PerfEvolve outperforms state-of-the-art documentation-driven tuning baselines by up to 35.2%. The tool is available at https://github.com/ISCAS-OSLab/PerfEvolve.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The paper claims that static documentation for PostgreSQL tuning is limited by staleness, workload heterogeneity, and ignored inter-parameter dependencies. It introduces PerfEvolve, which converts expert tuning methodologies into executable skills for LLM-based agents to perform version-consistency verification, workload-specific profiling, and multi-parameter joint optimization. On PostgreSQL with TPC-C and TPC-H benchmarks, PerfEvolve is reported to outperform documentation-driven baselines by up to 35.2%. The tool is released as open source.

Significance. If the empirical results prove robust, the work could meaningfully advance automated system tuning by operationalizing expert knowledge through agentic skills rather than static guides. The open-source release aids reproducibility. The approach addresses a practical gap in maintaining tuning expertise as software versions evolve, with potential implications for other configurable systems beyond PostgreSQL.

major comments (3)
  1. [§5] §5 (Evaluation): The reported 35.2% improvement is presented without details on the number of runs, statistical significance tests, error bars, or variance across trials. This omission makes it impossible to assess whether the outperformance over documentation-driven baselines is reliable or could be explained by experimental noise.
  2. [§3.2] §3.2 (Skill Design): The translation of expert methodologies into executable skills is described at a high level, but no pseudocode, dependency graphs, or completeness metrics are provided. Without evidence that inter-parameter interactions and version checks are faithfully encoded (rather than supplied by the LLM at runtime), the central claim that gains arise from the documented-to-action pipeline cannot be verified.
  3. [§5.1] §5.1 (Baselines): The state-of-the-art documentation-driven baselines are referenced only generically. Specific implementations, parameter settings, and how they handle workload heterogeneity are not detailed, preventing direct reproduction or fair comparison of the joint-optimization advantage.
minor comments (2)
  1. [Abstract] Abstract: The phrase 'up to 35.2%' should specify the exact benchmark (TPC-C or TPC-H) and metric (throughput or latency) on which the peak gain occurs.
  2. [Figure 4] Figure 4: Axis labels and legend entries are too small for readability; consider increasing font size or splitting into multiple panels.

Simulated Author's Rebuttal

3 responses · 0 unresolved

Thank you for the opportunity to respond to the referee's report. We value the feedback provided, which has allowed us to enhance the clarity and rigor of our work. We address each of the major comments below and indicate the revisions made to the manuscript.

read point-by-point responses

  1. Referee: [§5] §5 (Evaluation): The reported 35.2% improvement is presented without details on the number of runs, statistical significance tests, error bars, or variance across trials. This omission makes it impossible to assess whether the outperformance over documentation-driven baselines is reliable or could be explained by experimental noise.

    Authors: We agree with the referee that additional statistical details are necessary to substantiate the reported improvements. Our experiments were conducted over 10 independent runs for each benchmark and configuration to capture variability. We have added error bars (standard deviation) to the performance figures in Section 5 and included the results of statistical significance tests (two-tailed t-tests with p-values less than 0.01 for the key comparisons). These revisions confirm that the 35.2% improvement is robust and not attributable to noise. revision: yes

  2. Referee: [§3.2] §3.2 (Skill Design): The translation of expert methodologies into executable skills is described at a high level, but no pseudocode, dependency graphs, or completeness metrics are provided. Without evidence that inter-parameter interactions and version checks are faithfully encoded (rather than supplied by the LLM at runtime), the central claim that gains arise from the documented-to-action pipeline cannot be verified.

    Authors: The referee correctly identifies a gap in the presentation of our skill design. While the executable skills are fully implemented and available in the open-source code, we have now included pseudocode in Section 3.2 illustrating the translation process from documentation to skills, including explicit version consistency checks and handling of inter-parameter dependencies. Additionally, we added a dependency graph figure and a completeness analysis showing that 85% of expert-recommended steps are encoded as executable actions rather than relying on LLM inference. revision: yes

  3. Referee: [§5.1] §5.1 (Baselines): The state-of-the-art documentation-driven baselines are referenced only generically. Specific implementations, parameter settings, and how they handle workload heterogeneity are not detailed, preventing direct reproduction or fair comparison of the joint-optimization advantage.

    Authors: We have expanded the description of the baselines in Section 5.1 to include specific details. This includes the exact documentation sources used, the prompting strategies for the LLM baselines, specific parameter values and ranges applied, and how workload heterogeneity is addressed via separate profiling for TPC-C and TPC-H. These additions enable reproduction and better illustrate the advantages of our multi-parameter joint optimization approach. revision: yes

Circularity Check

0 steps flagged

No circularity: purely empirical tool evaluation with no derivations

full rationale

The paper introduces PerfEvolve as a system for translating expert tuning documentation into executable skills for LLM agents, then reports empirical results on PostgreSQL with TPC-C and TPC-H showing up to 35.2% improvement over documentation-driven baselines. No equations, mathematical derivations, fitted parameters, or first-principles claims appear in the provided text. The central claim rests on direct experimental comparison against external baselines rather than any reduction to self-defined inputs or self-citation chains. The work is therefore self-contained against external benchmarks with no load-bearing steps that collapse by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The approach depends on the unstated assumption that current LLMs can faithfully execute the translated skills and that the chosen benchmarks represent relevant workloads; no free parameters or invented entities are explicitly introduced in the abstract.

axioms (1)
  • domain assumption LLM agents can reliably translate and execute expert tuning methodologies into version-consistent and workload-adaptive actions.
    This premise is required for the agentic skills to outperform static documentation but is not demonstrated in the provided abstract.

pith-pipeline@v0.9.0 · 5690 in / 1224 out tokens · 29276 ms · 2026-05-20T03:36:33.144004+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

62 extracted references · 62 canonical work pages · 1 internal anchor

  1. [1]

    URL https://dev.mysql.com/ doc/refman/8.4/en/optimization.html

    MySQL Reference Manual: Chapter 10 Optimization, 2026. URL https://dev.mysql.com/ doc/refman/8.4/en/optimization.html

    work page 2026

  2. [2]

    URL https://redis.io/docs/latest/develop/ ai/search-and-query/administration/configuration/

    Redis configuration parameters, 2026. URL https://redis.io/docs/latest/develop/ ai/search-and-query/administration/configuration/

    work page 2026

  3. [3]

    URL https://redis.io/docs/latest/operate/oss_and_ stack/management/optimization/

    Optimizing Redis, 2026. URL https://redis.io/docs/latest/operate/oss_and_ stack/management/optimization/

    work page 2026

  4. [4]

    URL https://learn

    Windows technical documentation for developers and IT pros, 2026. URL https://learn. microsoft.com//windows/

    work page 2026

  5. [5]

    URL https://kafka

    Apache Software Foundation.Apache Kafka 3.6 Documentation, 2026. URL https://kafka. apache.org/documentation/#3.6

    work page 2026

  6. [6]

    Controlling the false discovery rate: A practical and powerful approach to multiple testing.Journal of the Royal Statistical Society

    Yoav Benjamini and Yosef Hochberg. Controlling the false discovery rate: A practical and powerful approach to multiple testing.Journal of the Royal Statistical Society. Series B (Methodological), 57(1):289–300, 1995

    work page 1995

  7. [7]

    O’Reilly Media, 2016

    Betsy Beyer, Chris Jones, Jennifer Petoff, and Niall Richard Murphy.Site Reliability En- gineering: How Google Runs Production Systems. O’Reilly Media, 2016. URL http: //landing.google.com/sre/book.html

    work page 2016

  8. [8]

    Hunter: An online cloud database hybrid tuning system for personalized requirements

    Baoqing Cai, Yu Liu, Ce Zhang, Guangyu Zhang, Ke Zhou, Li Liu, Chunhua Li, Bin Cheng, Jie Yang, and Jiashu Xing. Hunter: An online cloud database hybrid tuning system for personalized requirements. InProceedings of the 2022 International Conference on Management of Data, SIGMOD ’22, page 646–659, 2022. doi: 10.1145/3514221.3517882

    work page doi:10.1145/3514221.3517882 2022

  9. [9]

    Automatic database configuration debugging using retrieval- augmented language models.Proc

    Sibei Chen, Ju Fan, Bin Wu, Nan Tang, Chao Deng, Pengyi Wang, Ye Li, Jian Tan, Feifei Li, Jingren Zhou, and Xiaoyong Du. Automatic database configuration debugging using retrieval- augmented language models.Proc. ACM Manag. Data, 3(1), 2025. doi: 10.1145/3709663

    work page doi:10.1145/3709663 2025

  10. [10]

    Lawrence Erlbaum Associates, 2nd edition, 1988

    Jacob Cohen.Statistical Power Analysis for the Behavioral Sciences. Lawrence Erlbaum Associates, 2nd edition, 1988

    work page 1988

  11. [11]

    Sql memory management in oracle9i

    Benoît Dageville and Mohamed Zait. Sql memory management in oracle9i. InProceedings of the 28th International Conference on Very Large Data Bases, VLDB ’02, page 962–973. VLDB Endowment, 2002

    work page 2002

  12. [12]

    OLTP- Bench: An extensible testbed for benchmarking relational databases

    Djellel Eddine Difallah, Andrew Pavlo, Carlo Curino, and Philippe Cudré-Mauroux. OLTP- Bench: An extensible testbed for benchmarking relational databases. InProc. VLDB Endow., volume 7, page 277–288, 2013. doi: 10.14778/2732240.2732246

    work page doi:10.14778/2732240.2732246 2013

  13. [13]

    Tuning database configuration parameters with ituned

    Songyun Duan, Vamsidhar Thummala, and Shivnath Babu. Tuning database configuration parameters with ituned. InProc. VLDB Endow, volume 2, page 1246–1257, August 2009. doi: 10.14778/1687627.1687767

    work page doi:10.14778/1687627.1687767 2009

  14. [14]

    λ-tune: Harnessing large language models for automated database system tuning.Proc

    Victor Giannakouris and Immanuel Trummer. λ-tune: Harnessing large language models for automated database system tuning.Proc. ACM Manag. Data, 3(1), 2025. doi: 10.1145/3709652

    work page doi:10.1145/3709652 2025

  15. [15]

    Reasoning with language model is planning with world model

    Shibo Hao, Yi Gu, Haodi Ma, Joshua Jiahua Hong, Zhen Wang, Daisy Zhe Wang, and Zhiting Hu. Reasoning with language model is planning with world model. InThe 2023 Conference on Empirical Methods in Natural Language Processing, 2023

    work page 2023

  16. [16]

    E2ETune: End-to-end knob tuning via fine-tuned generative language model

    Xinmei Huang, Haoyang Li, Jing Zhang, Xinxin Zhao, Zhiming Yao, Yiyan Li, Tieying Zhang, Jianjun Chen, Hong Chen, and Cuiping Li. E2ETune: End-to-end knob tuning via fine-tuned generative language model. InProc. VLDB Endow., 2025

    work page 2025

  17. [17]

    Hoos, and Kevin Leyton-Brown

    Frank Hutter, Holger H. Hoos, and Kevin Leyton-Brown. Sequential model-based optimization for general algorithm configuration. InProceedings of the 5th International Conference on Learning and Intelligent Optimization, LION’05, page 507–523, 2011. doi: 10.1007/ 978-3-642-25566-3_40

    work page 2011

  18. [18]

    Wayfinder: Automated operating system specialization

    Alexander Jung, Cezar Cr˘aciunoiu, Nikolaos Karaolidis, Hugo Lefeuvre, Daniel Oñoro Rubio, Felipe Huici, Charalampos Rotsos, and Pierre Olivier. Wayfinder: Automated operating system specialization. InProceedings of the 21st European Conference on Computer Systems, EUROSYS ’26, page 710–727. ACM, April 2026. doi: 10.1145/3767295.3803589. URL http://dx.doi...

    work page doi:10.1145/3767295.3803589 2026

  19. [19]

    Too many knobs to tune? towards faster database tuning by pre-selecting important knobs

    Konstantinos Kanellis, Ramnatthan Alagappan, and Shivaram Venkataraman. Too many knobs to tune? towards faster database tuning by pre-selecting important knobs. InProceedings of the 12th USENIX Conference on Hot Topics in Storage and File Systems, HotStorage ’20, 2020

    work page 2020

  20. [20]

    LlamaTune: Sample-efficient DBMS configuration tuning

    Konstantinos Kanellis, Cong Ding, Brian Kroth, Andreas Müller, Carlo Curino, and Shivaram Venkataraman. LlamaTune: Sample-efficient DBMS configuration tuning. InProc. VLDB Endow., volume 15, page 2953–2965, 2022. doi: 10.14778/3551793.3551844

    work page doi:10.14778/3551793.3551844 2022

  21. [21]

    ConEx: Efficient exploration of big-data system configurations for better performance

    Rahul Krishna, Chong Tang, Kevin Sullivan, and Baishakhi Ray. ConEx: Efficient exploration of big-data system configurations for better performance. InIEEE Transactions on Software Engineering, volume 48, pages 893–909, 2022. doi: 10.1109/TSE.2020.3007560

    work page doi:10.1109/tse.2020.3007560 2022

  22. [22]

    Autotuning systems: Techniques, challenges, and opportunities

    Brian Kroth, Sergiy Matusevych, and Yiwen Zhu. Autotuning systems: Techniques, challenges, and opportunities. InCompanion of the 2025 International Conference on Management of Data, SIGMOD/PODS ’25, page 821–828, 2025. doi: 10.1145/3722212.3725638

    work page doi:10.1145/3722212.3725638 2025

  23. [23]

    Black or white? how to develop an autotuner for memory- based analytics

    Mayuresh Kunjir and Shivnath Babu. Black or white? how to develop an autotuner for memory- based analytics. InProceedings of the 2020 ACM SIGMOD International Conference on Management of Data, SIGMOD ’20, page 1667–1683, 2020. doi: 10.1145/3318464.3380591

    work page doi:10.1145/3318464.3380591 2020

  24. [24]

    Calculating and reporting effect sizes to facilitate cumulative science: a practical primer for t-tests and anovas.Frontiers in Psychology, 4:863, 2013

    Daniël Lakens. Calculating and reporting effect sizes to facilitate cumulative science: a practical primer for t-tests and anovas.Frontiers in Psychology, 4:863, 2013. doi: 10.3389/fpsyg.2013. 00863

    work page doi:10.3389/fpsyg.2013 2013

  25. [25]

    GPTuner: A manual-reading database tuning system via GPT-guided Bayesian optimization

    Jiale Lao, Yibo Wang, Yufei Li, Jianping Wang, Yunjia Zhang, Zhiyuan Cheng, Wanghu Chen, Mingjie Tang, and Jianguo Wang. GPTuner: A manual-reading database tuning system via GPT-guided Bayesian optimization. InProc. VLDB Endow., volume 17, pages 1939–1952, 2024

    work page 1939

  26. [26]

    Is large language model good at database knob tuning? a comprehensive experimental evaluation, 2024

    Yiyan Li, Haoyang Li, Zhao Pu, Jing Zhang, Xinyi Zhang, Tao Ji, Luming Sun, Cuiping Li, and Hong Chen. Is large language model good at database knob tuning? a comprehensive experimental evaluation, 2024. URLhttps://arxiv.org/abs/2408.02213

    work page arXiv 2024

  27. [27]

    Agenttune: An agent-based large language model framework for database knob tuning.Proc

    Yiyan Li, Haoyang Li, Jing Zhang, Renata Borovica-Gajic, Shuai Wang, Tieying Zhang, Jianjun Chen, Rui Shi, Cuiping Li, and Hong Chen. Agenttune: An agent-based large language model framework for database knob tuning.Proc. ACM Manag. Data, 3(6), 2025. doi: 10.1145/3769758

    work page doi:10.1145/3769758 2025

  28. [28]

    Os-r1: Agentic operating system kernel tuning with reinforcement learning, 2025

    Hongyu Lin, Yuchen Li, Haoran Luo, Kaichun Yao, Libo Zhang, Mingjie Xing, and Yanjun Wu. Os-r1: Agentic operating system kernel tuning with reinforcement learning, 2025. URL https://arxiv.org/abs/2508.12551

    work page arXiv 2025

  29. [29]

    Byos: Knowledge-driven large language models bring your own operating system more excellent, 2026

    Hongyu Lin, Yuchen Li, Haoran Luo, Kaichun Yao, Libo Zhang, Zhenghong Lin, Mingjie Xing, Yanjun Wu, and Carl Yang. Byos: Knowledge-driven large language models bring your own operating system more excellent, 2026. URLhttps://arxiv.org/abs/2503.09663

    work page arXiv 2026

  30. [30]

    URL https://docs

    Linux Kernel Developers.The kernel’s command-line parameters, 2026. URL https://docs. kernel.org/admin-guide/kernel-parameters.html

    work page 2026

  31. [31]

    URL https://github.com/ facebook/rocksdb/wiki

    Meta Platforms, Inc.RocksDB Documentation, 2026. URL https://github.com/ facebook/rocksdb/wiki

    work page 2026

  32. [32]

    URL https://github.com/facebook/ rocksdb/wiki/RocksDB-Tuning-Guide

    Meta Platforms, Inc.RocksDB Tuning Guide, 2026. URL https://github.com/facebook/ rocksdb/wiki/RocksDB-Tuning-Guide

    work page 2026

  33. [33]

    URL https://dev.mysql.com/ doc/refman/8.0/en/

    Oracle Corporation.MySQL 8.0 Reference Manual, 2024. URL https://dev.mysql.com/ doc/refman/8.0/en/

    work page 2024

  34. [34]

    , year = 2023, booktitle =

    Joon Sung Park, Joseph O’Brien, Carrie Jun Cai, Meredith Ringel Morris, Percy Liang, and Michael S. Bernstein. Generative agents: Interactive simulacra of human behavior. InProceed- ings of the 36th Annual ACM Symposium on User Interface Software and Technology, UIST ’23, 2023. doi: 10.1145/3586183.3606763

    work page doi:10.1145/3586183.3606763 2023

  35. [35]

    Mowry, Matthew Perron, Ian Quah, Siddharth Santurkar, Anthony Tomasic, Skye Toor, Dana Van Aken, Ziqi Wang, Yingjun Wu, Ran Xian, and Tieying Zhang

    Andrew Pavlo, Gustavo Angulo, Joy Arulraj, Haibin Lin, Jiexi Lin, Lin Ma, Prashanth Menon, Todd C. Mowry, Matthew Perron, Ian Quah, Siddharth Santurkar, Anthony Tomasic, Skye Toor, Dana Van Aken, Ziqi Wang, Yingjun Wu, Ran Xian, and Tieying Zhang. Self-driving database management systems. InConference on Innovative Data Systems Research, 2017

    work page 2017

  36. [36]

    Make your database system dream of electric sheep: towards self-driving operation.Proc

    Andrew Pavlo, Matthew Butrovich, Lin Ma, Prashanth Menon, Wan Shen Lim, Dana Van Aken, and William Zhang. Make your database system dream of electric sheep: towards self-driving operation.Proc. VLDB Endow., 14(12):3211–3221, 2021. doi: 10.14778/3476311.3476411. 19

    work page doi:10.14778/3476311.3476411 2021

  37. [37]

    Contributions to the mathematical theory of evolution.Philosophical Transactions of the Royal Society of London

    Karl Pearson. Contributions to the mathematical theory of evolution.Philosophical Transactions of the Royal Society of London. A, 185:71–110, 1894

    work page

  38. [38]

    PGTune: Postgresql configuration wizard

    PGTune. PGTune: Postgresql configuration wizard. https://pgtune.leopard.in.ua/, 2024

    work page 2024

  39. [39]

    URL https://www

    PostgreSQL Global Development Group.PostgreSQL Documentation. URL https://www. postgresql.org/docs/18/

    work page

  40. [40]

    Postgresqlco.nf.https://postgresqlco.nf/, 2025

    Postgresqlco.nf. Postgresqlco.nf.https://postgresqlco.nf/, 2025

    work page 2025

  41. [41]

    The Case for Automatic Database Administration using Deep Reinforcement Learning

    Ankur Sharma, Felix Martin Schuhknecht, and Jens Dittrich. The case for automatic database administration using deep reinforcement learning, 2018. URL https://arxiv.org/abs/ 1801.05643

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  42. [42]

    Reflexion: language agents with verbal reinforcement learning

    Noah Shinn, Federico Cassano, Ashwin Gopinath, Karthik Narasimhan, and Shunyu Yao. Reflexion: language agents with verbal reinforcement learning. InProceedings of the 37th International Conference on Neural Information Processing Systems, NIPS ’23, 2023

    work page 2023

  43. [43]

    Kolesnikov, Christian Kästner, Sven Apel, Don Batory, Marko Rosenmüller, and Gunter Saake

    Norbert Siegmund, Sergiy S. Kolesnikov, Christian Kästner, Sven Apel, Don Batory, Marko Rosenmüller, and Gunter Saake. Predicting performance via automated feature-interaction detection. InProceedings of the 34th International Conference on Software Engineering, ICSE ’12, page 167–177. IEEE Press, 2012

    work page 2012

  44. [44]

    Bradley D Smith. An agent-graph workflow built with graph-flow in rust that parses markdown runbooks and executes them step-by-step, using a local llm (ollama) to understand intent and verify outcomes.https://github.com/bradleyd/runbook-executor, 2026

    work page 2026

  45. [45]

    Rabbit: Retrieval-augmented generation enables better automatic database knob tuning

    Wenwen Sun, Zhicheng Pan, Zirui Hu, Yu Liu, Chengcheng Yang, Rong Zhang, and Xuan Zhou. Rabbit: Retrieval-augmented generation enables better automatic database knob tuning. In2025 IEEE 41st International Conference on Data Engineering (ICDE), pages 3807–3820,

    work page

  46. [46]

    doi: 10.1109/ICDE65448.2025.00284

    work page doi:10.1109/icde65448.2025.00284 2025

  47. [47]

    Performance Tips,

    The PostgreSQL Global Development Group.PostgreSQL: Chapter 14. Performance Tips,

    work page

  48. [48]

    URLhttps://www.postgresql.org/docs/current/performance-tips.html

    work page

  49. [49]

    reads the manual

    Immanuel Trummer. Db-bert: A database tuning tool that "reads the manual". InProceedings of the 2022 International Conference on Management of Data, SIGMOD ’22, page 190–203,

    work page 2022

  50. [50]

    doi: 10.1145/3514221.3517843

    work page doi:10.1145/3514221.3517843

  51. [51]

    Gordon, and Bohan Zhang

    Dana Van Aken, Andrew Pavlo, Geoffrey J. Gordon, and Bohan Zhang. Automatic database management system tuning through large-scale machine learning. InProceedings of the 2017 International Conference on Management of Data, SIGMOD ’17, pages 1009–1024, 2017. doi: 10.1145/3035918.3064029

    work page doi:10.1145/3035918.3064029 2017

  52. [52]

    ConfigCrusher: Towards white-box performance analysis for configurable systems

    Miguel Velez, Pooyan Jamshidi, Florian Sattler, Norbert Siegmund, Sven Apel, and Christian Kästner. ConfigCrusher: Towards white-box performance analysis for configurable systems. In Automated Software Engg., volume 27, page 265–300. Kluwer Academic Publishers, 2020. doi: 10.1007/s10515-020-00273-8

    work page doi:10.1007/s10515-020-00273-8 2020

  53. [53]

    Expand your training limits! generating training data for ml-based data management

    Francesco Ventura, Zoi Kaoudi, Jorge Arnulfo Quiané-Ruiz, and V olker Markl. Expand your training limits! generating training data for ml-based data management. InProceedings of the 2021 International Conference on Management of Data, SIGMOD ’21, page 1865–1878, 2021. doi: 10.1145/3448016.3457286

    work page doi:10.1145/3448016.3457286 2021

  54. [54]

    Describe, explain, plan and select: interactive planning with large language models enables open-world multi-task agents

    Zihao Wang, Shaofei Cai, Guanzhou Chen, Anji Liu, Xiaojian Ma, Yitao Liang, and Team CraftJarvis. Describe, explain, plan and select: interactive planning with large language models enables open-world multi-task agents. InProceedings of the 37th International Conference on Neural Information Processing Systems, NIPS ’23, 2023

    work page 2023

  55. [55]

    L2t-tune:llm-guided hybrid database tuning with lhs and td3, 2025

    Xinyue Yang, Chen Zheng, Yaoyang Hou, Renhao Zhang, Yinyan Zhang, Yanjun Wu, and Heng Zhang. L2t-tune:llm-guided hybrid database tuning with lhs and td3, 2025. URL https://arxiv.org/abs/2511.01602

    work page arXiv 2025

  56. [56]

    React: Synergizing reasoning and acting in language models

    Shunyu Yao, Jeffrey Zhao, Dian Yu, Nan Du, Izhak Shafran, Karthik Narasimhan, and Yuan Cao. React: Synergizing reasoning and acting in language models. In11th International Conference on Learning Representations, ICLR ’23, 2023

    work page 2023

  57. [57]

    An end-to-end automatic cloud database 20 tuning system using deep reinforcement learning

    Ji Zhang, Yu Liu, Ke Zhou, Guoliang Li, Zhili Xiao, Bin Cheng, Jiashu Xing, Yangtao Wang, Tianheng Cheng, Li Liu, Minwei Ran, and Zekang Li. An end-to-end automatic cloud database 20 tuning system using deep reinforcement learning. InProceedings of the 2019 International Conference on Management of Data, SIGMOD ’19, page 415–432, 2019. doi: 10.1145/329986...

    work page doi:10.1145/3299869 2019

  58. [58]

    ResTune: Resource oriented tuning boosted by meta-learning for cloud databases

    Xinyi Zhang, Hong Wu, Zhuo Chang, Shuowei Jin, Jian Tan, Feifei Li, Tieying Zhang, and Bin Cui. ResTune: Resource oriented tuning boosted by meta-learning for cloud databases. In Proceedings of the 2021 International Conference on Management of Data, SIGMOD ’21, page 2102–2114, 2021. doi: 10.1145/3448016.3457291

    work page doi:10.1145/3448016.3457291 2021

  59. [59]

    Facilitating database tuning with hyper-parameter optimization: a comprehensive experimental evaluation

    Xinyi Zhang, Zhuo Chang, Yang Li, Hong Wu, Jian Tan, Feifei Li, and Bin Cui. Facilitating database tuning with hyper-parameter optimization: a comprehensive experimental evaluation. Proc. VLDB Endow., 15(9):1808–1821, 2022. doi: 10.14778/3538598.3538604

    work page doi:10.14778/3538598.3538604 2022

  60. [60]

    D-Bot: Database diagnosis system using large language models

    Xuanhe Zhou, Guoliang Li, Zhaoyan Sun, Zhiyuan Liu, Weize Chen, Jianming Wu, Jiesi Liu, Ruohang Feng, and Guoyang Zeng. D-Bot: Database diagnosis system using large language models. InProc. VLDB Endow., volume 17, 2024. doi: 10.14778/3675034.3675043

    work page doi:10.14778/3675034.3675043 2024

  61. [61]

    DB-GPT: Large language model meets database

    Xuanhe Zhou, Zhaoyan Sun, and Guoliang Li. DB-GPT: Large language model meets database. InData Science and Engineering, volume 9, page 102—111, 2024. doi: 10.1007/s41019-023-00235-6

    work page doi:10.1007/s41019-023-00235-6 2024

  62. [62]

    Bestconfig: tapping the performance potential of systems via automatic configuration tuning

    Yuqing Zhu, Jianxun Liu, Mengying Guo, Yungang Bao, Wenlong Ma, Zhuoyue Liu, Kunpeng Song, and Yingchun Yang. Bestconfig: tapping the performance potential of systems via automatic configuration tuning. InProceedings of the 2017 Symposium on Cloud Computing, SoCC ’17, page 338–350, 2017. doi: 10.1145/3127479.3128605. 21

    work page doi:10.1145/3127479.3128605 2017