pith. machine review for the scientific record. sign in

arxiv: 2605.02377 · v1 · submitted 2026-05-04 · 💻 cs.DB

Recognition: unknown

Unfair by design: eBPF-based scheduling of mixed database workloads

Bettina Kemme, Carl-Elliott Bilodeau-Savaria, Jan Kristof Nidzwetzki, Stefanie Scherzinger

Pith reviewed 2026-05-08 02:17 UTC · model grok-4.3

classification 💻 cs.DB
keywords eBPF schedulingmixed database workloadspriority inversiontail latencyPostgreSQLLinux kernelbackground tasksimmediate preemption
0
0 comments X

The pith

UFS uses eBPF to let background database tasks run only on idle CPUs while immediately preempting them for time-sensitive work and sharing lock hints to avoid priority inversion.

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

Database systems increasingly mix latency-critical queries with background tasks that should consume only spare resources. Existing Linux schedulers fail to isolate the critical work reliably and allow priority inversion when background tasks hold locks. The paper introduces UFS, an eBPF sched_ext scheduler that restricts background tasks to unused CPU capacity, preempts them at once when critical tasks arrive, and accepts application hints via eBPF maps so background tasks release locks promptly. Integration into PostgreSQL under mixed workloads produces up to 2X higher throughput and half the tail latency for time-sensitive tasks compared with standard Linux options.

Core claim

UFS is a selectively unfair scheduler built as an eBPF-based sched_ext extension in the Linux kernel. It limits background tasks to idle CPU capacity and enforces immediate preemption for arriving time-sensitive tasks. Application-level lock hints are passed through eBPF maps so that background tasks holding locks do not unnecessarily delay waiting time-sensitive work. When integrated with PostgreSQL, the approach yields up to 2X throughput gains and a 50 percent reduction in tail latency for time-sensitive tasks under mixed workloads versus existing Linux schedulers.

What carries the argument

UFS, the eBPF sched_ext scheduler that enforces immediate preemption of background tasks and incorporates application lock hints via eBPF maps to prevent priority inversion.

If this is right

  • Time-sensitive database tasks achieve up to 2X higher throughput under mixed loads.
  • Tail latency for time-sensitive tasks drops by half compared with standard Linux schedulers.
  • Priority inversion is reduced because background tasks receive hints to release locks quickly.
  • Background tasks consume only spare CPU capacity and do not starve critical work.

Where Pith is reading between the lines

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

  • The same immediate-preemption-plus-hints pattern could be applied to other systems that mix request handling with batch or maintenance work.
  • Wider availability of sched_ext may reduce the need for databases to implement complex internal task isolation logic.
  • Further measurements could check whether the added eBPF path introduces measurable overhead at very high concurrency.

Load-bearing premise

The eBPF sched_ext implementation can reliably enforce immediate preemption and correctly interpret application-level lock hints without introducing unacceptable kernel overhead or stability issues.

What would settle it

A controlled run in which a time-sensitive task becomes ready while a background task holds a contended lock, followed by measurement of whether the time-sensitive task is delayed beyond normal lock-wait time or whether preemption occurs within one scheduling quantum.

Figures

Figures reproduced from arXiv: 2605.02377 by Bettina Kemme, Carl-Elliott Bilodeau-Savaria, Jan Kristof Nidzwetzki, Stefanie Scherzinger.

Figure 2
Figure 2. Figure 2: CPU utilization by high￾prio workers across 4 CPUs view at source ↗
Figure 4
Figure 4. Figure 4: Enqueue strategies. the current PostgreSQL backend to the appropriate cgroup based on SQL configuration variables. Assignment can be done dynamically. Task Scheduling. Tasks are scheduled according to their tier w.r.t. (1) priority under contention: which task is allowed to run when CPU time is scarce, and (2) CPU selection and dispatch: how a task is assigned to a CPU and how it reaches that CPU’s run que… view at source ↗
Figure 5
Figure 5. Figure 5: Each node in this tree represents a cgroup DSQ, using its view at source ↗
Figure 6
Figure 6. Figure 6: Throughput for CPU-bound vs. CPU-bursty tasks view at source ↗
Figure 8
Figure 8. Figure 8: Throughput under different priorities for CPU view at source ↗
Figure 7
Figure 7. Figure 7: Scaling throughput for CPU-bursty tasks for 8/16/24 view at source ↗
Figure 9
Figure 9. Figure 9: Benchmarking schedulers under general workloads, view at source ↗
Figure 10
Figure 10. Figure 10: Throughput for a workload mix of CPU-bursty view at source ↗
read the original abstract

Modern database systems increasingly co-schedule time-sensitive and background tasks. In such mixed workloads, background tasks should ideally utilize only spare CPU capacity without interfering with latency-critical requests. While some database-level solutions address this challenge, many database systems still rely on operating system (OS) schedulers, which, despite supporting priorities, do not reliably isolate high-priority tasks. Furthermore, they remain vulnerable to priority inversion, where preempted background tasks can delay other work. We present UFS, a selectively unfair scheduler implemented as an eBPF-based sched_ext scheduler in the Linux kernel. UFS restricts background tasks to idle CPU capacity and preempts them immediately when time-sensitive tasks arrive. To address priority inversion, UFS incorporates application-level hints via eBPF maps, ensuring that background tasks are not unnecessarily delayed should time-sensitive tasks wait for them to release locks. Our integration of UFS into PostgreSQL demonstrates that, under mixed workloads, UFS improves throughput for time-sensitive tasks by up to 2X, while reducing tail latency by half, compared to existing scheduling options in Linux.

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

2 major / 1 minor

Summary. The paper introduces UFS, a selectively unfair scheduler implemented as an eBPF-based sched_ext scheduler in the Linux kernel. It aims to restrict background tasks to idle CPU capacity, preempt them immediately for time-sensitive tasks, and use application-level hints via eBPF maps to avoid priority inversion. The integration with PostgreSQL is reported to improve throughput for time-sensitive tasks by up to 2X and reduce tail latency by half under mixed workloads compared to existing Linux scheduling options.

Significance. If the results are confirmed with rigorous experiments, this could be significant for improving workload isolation in database systems using OS-level scheduling enhancements. The eBPF approach allows for flexible, application-aware scheduling without modifying the database code extensively, potentially influencing future OS-DB co-design.

major comments (2)
  1. Abstract: The headline claims of up to 2X throughput improvement and 50% tail latency reduction are presented without any experimental details, workload definitions, baseline configurations, preemption-latency numbers, or eBPF overhead breakdowns, making it impossible to judge whether the data support the stated mechanism.
  2. Implementation: The description of immediate preemption and lock-hint handling via eBPF maps does not include any measurements of decision latency, map access costs, or stability under PostgreSQL lock contention, which are required to confirm that sched_ext constraints are met and that gains can be attributed to UFS rather than other factors.
minor comments (1)
  1. Define acronyms such as UFS and sched_ext on first use for clarity.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. We address each major comment point by point below and have revised the paper to strengthen the presentation of results and implementation details.

read point-by-point responses

  1. Referee: Abstract: The headline claims of up to 2X throughput improvement and 50% tail latency reduction are presented without any experimental details, workload definitions, baseline configurations, preemption-latency numbers, or eBPF overhead breakdowns, making it impossible to judge whether the data support the stated mechanism.

    Authors: We agree that the abstract is high-level by design and does not embed the full experimental parameters. To improve verifiability while respecting length constraints, we will revise the abstract to briefly note the workload (mixed PostgreSQL time-sensitive OLTP queries and background tasks), the baselines (Linux CFS and priority schedulers), and that measurements are from controlled experiments with details in Sections 4-5. Preemption latency and eBPF overheads are quantified in the evaluation; we will add a forward reference in the abstract. revision: yes

  2. Referee: Implementation: The description of immediate preemption and lock-hint handling via eBPF maps does not include any measurements of decision latency, map access costs, or stability under PostgreSQL lock contention, which are required to confirm that sched_ext constraints are met and that gains can be attributed to UFS rather than other factors.

    Authors: The implementation section describes the sched_ext-based immediate preemption and eBPF map hints for lock handling, but we acknowledge the absence of dedicated microbenchmark numbers for decision latency, map costs, and contention stability. In the revised manuscript we will add a dedicated microbenchmark subsection to the evaluation, reporting sched_ext decision latency (sub-microsecond), eBPF map access overhead, and stability under PostgreSQL lock contention. These measurements will help confirm that observed gains are due to UFS and that sched_ext constraints are satisfied. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical implementation paper with no derivation chain

full rationale

The paper presents the design, eBPF implementation, and PostgreSQL integration of UFS, followed by empirical throughput and latency measurements under mixed workloads. No equations, first-principles derivations, fitted parameters, or predictions appear in the abstract or described content. Claims rest on direct experimental comparison to Linux schedulers rather than any self-referential reduction, self-citation load-bearing argument, or ansatz smuggled via prior work. This is a standard systems evaluation paper whose central results are falsifiable by replication and do not reduce to their inputs by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

The central claim rests on the unverified assumption that eBPF sched_ext can implement the described preemption and hint-passing behavior without side effects; no free parameters or invented physical entities are introduced.

axioms (1)
  • domain assumption eBPF sched_ext can implement immediate preemption and map-based hints for lock awareness
    Invoked implicitly when the paper states that UFS restricts background tasks and uses eBPF maps to avoid priority inversion.
invented entities (1)
  • UFS scheduler no independent evidence
    purpose: Selectively unfair scheduling for mixed database workloads
    New kernel component introduced by the paper; no independent evidence outside the implementation is provided.

pith-pipeline@v0.9.0 · 5503 in / 1273 out tokens · 88539 ms · 2026-05-08T02:17:08.607852+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

67 extracted references · 36 canonical work pages

  1. [1]

    Mumtaz Ahmad, Ashraf Aboulnaga, Shivnath Babu, and Kamesh Munagala

  2. [2]

    In2008 IEEE 24th International Conference on Data Engineering

    QShuffler: Getting the Query Mix Right. In2008 IEEE 24th International Conference on Data Engineering. 1415–1417. doi:10.1109/ICDE.2008.4497574

    work page doi:10.1109/icde.2008.4497574 2008

  3. [3]

    Matthew Butrovich, Samuel Arch, Wan Shen Lim, William Zhang, Jignesh M Patel, and Andrew Pavlo. 2025. BPF-DB: A Kernel-Embedded Transactional Database Management System For eBPF Applications.Proc. ACM Manag. Data 3, 3, Article 135 (June 2025), 27 pages. doi:10.1145/3725272

    work page doi:10.1145/3725272 2025

  4. [4]

    Matthew Butrovich, Karthik Ramanathan, John Rollinson, Wan Shen Lim, William Zhang, Justine Sherry, and Andrew Pavlo. 2023. Tigger: A Database Proxy That Bounces with User-Bypass.Proc. VLDB Endow.16, 11 (July 2023), 3335–3348. doi:10.14778/3611479.3611530

    work page doi:10.14778/3611479.3611530 2023

  5. [5]

    Carey, Rajiv Jauhari, and Miron Livny

    Michael J. Carey, Rajiv Jauhari, and Miron Livny. 1989. Priority in DBMS Resource Scheduling. InProceedings of the Fifteenth International Conference on Very Large Data Bases, August 22-25, 1989, Amsterdam, The Netherlands, Peter M. G. Apers and Gio Wiederhold (Eds.). Morgan Kaufmann, 397–410. http://www.vldb.org/ conf/1989/P397.PDF

    1989

  6. [6]

    Sean Wang

    Audrey Cheng, Aaron N. Kabcenell, Jason Chan, Xiao Shi, Peter D. Bailis, Natacha Crooks, and Ion Stoica. 2024. Towards Optimal Transaction Scheduling.Proc. VLDB Endow.17, 11 (2024), 2694–2707. doi:10.14778/3681954.3681956 Unfair by design: eBPF-based scheduling of mixed database workloads

    work page doi:10.14778/3681954.3681956 2024

  7. [7]

    Daniel Bristot de Oliveira. 2023. DL Server. Linux Plumbers Conference

    2023

  8. [8]

    https://lpc.events/event/17/contributions/1491/attachments/1239/2519/ DL%20Server%20LPC%202023.pdf, Accessed: 2026-04-21

    2026

  9. [9]

    Mathieu Desnoyers. 2023. [RFC PATCH] sched/fair: Bias runqueue selection towards almost idle prev CPU. https://lore.kernel.org/lkml/20230905171105. 1005672-2-mathieu.desnoyers@efficios.com/

    work page arXiv 2023

  10. [10]

    Djellel Eddine Difallah, Andrew Pavlo, Carlo Curino, and Philippe Cudré- Mauroux. 2013. OLTP-Bench: An Extensible Testbed for Benchmarking Re- lational Databases.PVLDB7, 4 (2013), 277–288. http://www.vldb.org/pvldb/ vol7/p277-difallah.pdf

    2013

  11. [11]

    Joseph Vinish D’silva, Florestan De Moor, and Bettina Kemme. 2018. AIDA: Abstraction for Advanced in-Database Analytics.Proc. VLDB Endow.11, 11 (jul 2018), 1400–1413. doi:10.14778/3236187.3236194

    work page doi:10.14778/3236187.3236194 2018

  12. [12]

    Kayhan Dursun, Carsten Binnig, Ugur Çetintemel, Garret Swart, and Weiwei Gong. 2019. A Morsel-Driven Query Execution Engine for Heterogeneous Multi- Cores.Proc. VLDB Endow.12, 12 (2019), 2218–2229. doi:10.14778/3352063.3352137

    work page doi:10.14778/3352063.3352137 2019

  13. [13]

    Yannis Foufoulas, Theoni Palaiologou, and Alkis Simitsis. 2025. The UDFBench Benchmark for General-purpose UDF Queries.Proc. VLDB Endow.18, 9 (2025), 2804–2817. doi:10.14778/3746405.3746409

    work page doi:10.14778/3746405.3746409 2025

  14. [14]

    Yannis Foufoulas and Alkis Simitsis. 2023. Efficient Execution of User-Defined Functions in SQL Queries.Proc. VLDB Endow.16, 12 (2023), 3874–3877. doi:10. 14778/3611540.3611574

    work page arXiv 2023

  15. [15]

    Joshua Fried, Zhenyuan Ruan, Amy Ousterhout, and Adam Belay. 2020. Caladan: Mitigating Interference at Microsecond Timescales. In14th USENIX Sympo- sium on Operating Systems Design and Implementation, OSDI 2020, Virtual Event, November 4-6, 2020. USENIX Association, 281–297. https://www.usenix.org/ conference/osdi20/presentation/fried

    2020

  16. [16]

    Yongjun He, Jiacheng Lu, and Tianzheng Wang. 2020. CoroBase: Coroutine- Oriented Main-Memory Database Engine.Proc. VLDB Endow.14, 3 (2020), 431–

    2020

  17. [17]

    doi:10.5555/3430915.3442440

    work page doi:10.5555/3430915.3442440

  18. [18]

    Hellerstein, Christoper Ré, Florian Schoppmann, Daisy Zhe Wang, Eugene Fratkin, Aleksander Gorajek, Kee Siong Ng, Caleb Welton, Xixuan Feng, Kun Li, and Arun Kumar

    Joseph M. Hellerstein, Christoper Ré, Florian Schoppmann, Daisy Zhe Wang, Eugene Fratkin, Aleksander Gorajek, Kee Siong Ng, Caleb Welton, Xixuan Feng, Kun Li, and Arun Kumar. 2012. The MADlib Analytics Library: Or MAD Skills, the SQL.Proc. VLDB Endow.5, 12 (aug 2012), 1700–1711. doi:10.14778/2367502. 2367510

    work page doi:10.14778/2367502 2012

  19. [19]

    Kaisong Huang, Tianzheng Wang, Qingqing Zhou, and Qingzhong Meng. 2023. The Art of Latency Hiding in Modern Database Engines.Proc. VLDB Endow.17, 3 (2023), 577–590. doi:10.14778/3632093.3632117

    work page doi:10.14778/3632093.3632117 2023

  20. [20]

    Kaisong Huang, Jiatang Zhou, Zhuoyue Zhao, Dong Xie, and Tianzheng Wang

  21. [21]

    ACM Manag

    Low-Latency Transaction Scheduling via Userspace Interrupts: Why Wait or Yield When You Can Preempt?Proc. ACM Manag. Data3, 3 (2025), 182:1– 182:25. doi:10.1145/3725319

    work page doi:10.1145/3725319 2025

  22. [22]

    2008.DB2 Workload Manager for Linux, UNIX, and Windows

    IBM Redbooks. 2008.DB2 Workload Manager for Linux, UNIX, and Windows. IBM Corporation, Poughkeepsie, NY, USA. https://www.redbooks.ibm.com/ abstracts/sg247524.html

    2008

  23. [23]

    2023.IBM Db2 13 for z/OS Performance Topics

    IBM Redbooks. 2023.IBM Db2 13 for z/OS Performance Topics. Number SG24- 8536-00 in IBM Redbooks. IBM Corporation, Poughkeepsie, NY, USA. https: //www.redbooks.ibm.com/redbooks/pdfs/sg248536.pdf IBM Form Number: SG24- 8536-00

    2023

  24. [24]

    Kostis Kaffes, Timothy Chong, Jack Tigar Humphries, Adam Belay, David Mazières, and Christos Kozyrakis. 2019. Shinjuku: Preemptive Scheduling for 𝜇second-scale Tail Latency. In16th USENIX Symposium on Networked Sys- tems Design and Implementation, NSDI 2019, Boston, MA, February 26-28, 2019, Jay R. Lorch and Minlan Yu (Eds.). USENIX Association, 345–360. ...

    2019

  25. [25]

    Kangnyeon Kim, Tianzheng Wang, Ryan Johnson, and Ippokratis Pandis. 2016. ERMIA: Fast Memory-Optimized Database System for Heterogeneous Work- loads. InProceedings of the 2016 International Conference on Management of Data, SIGMOD Conference 2016, San Francisco, CA, USA, June 26 - July 01, 2016, Fatma Özcan, Georgia Koutrika, and Sam Madden (Eds.). ACM, 1...

    work page doi:10.1145/2882903.2882905 2016

  26. [26]

    Viktor Leis, Peter Boncz, Alfons Kemper, and Thomas Neumann. 2014. Morsel- driven parallelism: a NUMA-aware query evaluation framework for the many- core age. InInternational Conference on Management of Data, SIGMOD 2014, Snowbird, UT, USA, June 22-27, 2014, Curtis E. Dyreson, Feifei Li, and M. Tamer Özsu (Eds.). ACM, 743–754. doi:10.1145/2588555.2610507

    work page doi:10.1145/2588555.2610507 2014

  27. [27]

    Viktor Leis and Christian Dietrich. 2024. Cloud-Native Database Systems and Unikernels: Reimagining OS Abstractions for Modern Hardware.Proc. VLDB Endow.17, 8 (2024), 2115–2122. doi:10.14778/3659437.3659462

    work page doi:10.14778/3659437.3659462 2024

  28. [28]

    Levandoski, David B

    Justin J. Levandoski, David B. Lomet, Sudipta Sengupta, Ryan Stutsman, and Rui Wang. 2015. Multi-Version Range Concurrency Control in Deuteronomy.Proc. VLDB Endow.8, 13 (2015), 2146–2157. doi:10.14778/2831360.2831368

    work page doi:10.14778/2831360.2831368 2015

  29. [29]

    Jacob Leverich and Christos Kozyrakis. 2014. Reconciling high server utilization and sub-millisecond quality-of-service. InNinth Eurosys Conference 2014, Eu- roSys 2014, Amsterdam, The Netherlands, April 13-16, 2014, Dick C. A. Bulterman, Herbert Bos, Antony I. T. Rowstron, and Peter Druschel (Eds.). ACM, 4:1–4:14. doi:10.1145/2592798.2592821

    work page doi:10.1145/2592798.2592821 2014

  30. [30]

    Edward Suh, Kostis Kaffes, and Christina Delimitrou

    Yueying Li, Nikita Lazarev, David Koufaty, Tenny Yin, Andy Anderson, Zhiru Zhang, G. Edward Suh, Kostis Kaffes, and Christina Delimitrou. 2024. LibPre- emptible: Enabling Fast, Adaptive, and Hardware-Assisted User-Space Sched- uling. InIEEE International Symposium on High-Performance Computer Archi- tecture, HPCA 2024, Edinburgh, United Kingdom, March 2-6...

    work page doi:10.1109/hpca57654.2024.00075 2024

  31. [31]

    Andersen

    Hyeontaek Lim, Michael Kaminsky, and David G. Andersen. 2017. Cicada: Dependably Fast Multi-Core In-Memory Transactions. InProceedings of the 2017 ACM International Conference on Management of Data, SIGMOD Conference 2017, Chicago, IL, USA, May 14-19, 2017, Semih Salihoglu, Wenchao Zhou, Rada Chirkova, Jun Yang, and Dan Suciu (Eds.). ACM, 21–35. doi:10.11...

    work page doi:10.1145/3035918 2017

  32. [32]

    Qiuru Lin, Sai Wu, Junbo Zhao, Jian Dai, Feifei Li, and Gang Chen. 2022. A Com- parative Study of in-Database Inference Approaches. In38th IEEE International Conference on Data Engineering, ICDE 2022, Kuala Lumpur, Malaysia, May 9-12,

    2022

  33. [33]

    doi:10.1109/ICDE53745.2022.00180

    IEEE, 1794–1807. doi:10.1109/ICDE53745.2022.00180

    work page doi:10.1109/icde53745.2022.00180 2022

  34. [34]

    Linux kernel documentation. [n. d.]. Extensible Scheduler Class (sched_ext). https://docs.kernel.org/scheduler/sched-ext.html. Accessed 2026-01-04

    2026

  35. [35]

    Linux kernel documentation. 2026. EEVDF Scheduler. https://docs.kernel.org/ scheduler/sched-eevdf.html. Accessed 2026-05-01

    2026

  36. [36]

    2016.wastedcores: Important note about the patches

    Jean-Pierre Lozi. 2016.wastedcores: Important note about the patches. https: //github.com/jplozi/wastedcores Repository README section stating the patches are not intended as generic bug fixes

    2016

  37. [37]

    Funston, Fabien Gaud, Vivien Quéma, and Alexandra Fedorova

    Jean-Pierre Lozi, Baptiste Lepers, Justin R. Funston, Fabien Gaud, Vivien Quéma, and Alexandra Fedorova. 2016. The Linux scheduler: a decade of wasted cores. InProceedings of the Eleventh European Conference on Computer Systems, EuroSys 2016, London, United Kingdom, April 18-21, 2016. ACM, 1:1–1:16. doi:10.1145/ 2901318.2901326

    work page arXiv 2016

  38. [38]

    Chris Mason. 2016. schbench: Scheduler Benchmark. https://kernel.googlesource. com/pub/scm/linux/kernel/git/mason/schbench. Accessed: 2026-04-10. A sched- uler benchmark that measures wakeup latency and throughput

    2016

  39. [39]

    Sarah McClure, Amy Ousterhout, Scott Shenker, and Sylvia Ratnasamy. 2022. Efficient Scheduling Policies for Microsecond-Scale Tasks. In19th USENIX Sym- posium on Networked Systems Design and Implementation, NSDI 2022, Renton, W A, USA, April 4-6, 2022, Amar Phanishayee and Vyas Sekar (Eds.). USENIX Associa- tion, 1–18. https://www.usenix.org/conference/ns...

    2022

  40. [40]

    McWherter, Bianca Schroeder, Anastassia Ailamaki, and Mor Harchol- Balter

    David T. McWherter, Bianca Schroeder, Anastassia Ailamaki, and Mor Harchol- Balter. 2005. Improving Preemptive Prioritization via Statistical Characterization of OLTP Locking. InProceedings of the 21st International Conference on Data Engineering, ICDE 2005, 5-8 April 2005, Tokyo, Japan, Karl Aberer, Michael J. Franklin, and Shojiro Nishio (Eds.). IEEE Co...

    2005

  41. [41]

    Oracle Corporation. [n. d.].Oracle Database 19c: Managing Resources with Oracle Database Resource Manager. Oracle Corporation. https: //docs.oracle.com/en/database/oracle/oracle-database/19/admin/managing- resources-with-oracle-database-resource-manager.html Accessed 2026-03-20

    2026

  42. [42]

    Matteo Paganelli, Paolo Sottovia, Kwanghyun Park, Matteo Interlandi, and Francesco Guerra. 2023. Pushing ML Predictions Into DBMSs.IEEE Trans. Knowl. Data Eng.35, 10 (2023), 10295–10308. doi:10.1109/TKDE.2023.3269592

    work page doi:10.1109/tkde.2023.3269592 2023

  43. [43]

    Sarvesh Pandey and Udai Shanker. 2020. Transaction scheduling protocols for controlling priority inversion: A review.Comput. Sci. Rev.35 (2020), 100215. doi:10.1016/J.COSREV.2019.100215

    work page doi:10.1016/j.cosrev.2019.100215 2020

  44. [44]

    The Linux Kernel Documentation Project. [n. d.].BPF Documentation. https: //docs.kernel.org/bpf/index.html Accessed: 2026-01-11

    2026

  45. [45]

    Hanson, Robert Walzer, Rodrigo Gomes, and Nikita Shamgunov

    Adam Prout, Szu-Po Wang, Joseph Victor, Zhou Sun, Yongzhu Li, Jack Chen, Evan Bergeron, Eric N. Hanson, Robert Walzer, Rodrigo Gomes, and Nikita Shamgunov. 2022. Cloud-Native Transactions and Analytics in SingleStore. In SIGMOD ’22: International Conference on Management of Data, Philadelphia, PA, USA, June 12 - 17, 2022, Zachary G. Ives, Angela Bonifati,...

    work page doi:10.1145/3514221.3526055 2022

  46. [46]

    Iraklis Psaroudakis, Tobias Scheuer, Norman May, Abdelkader Sellami, and Anas- tasia Ailamaki. 2016. Adaptive NUMA-aware data placement and task scheduling for analytical workloads in main-memory column-stores.Proc. VLDB Endow.10, 2 (2016), 37–48. doi:10.14778/3015274.3015275

    work page doi:10.14778/3015274.3015275 2016

  47. [47]

    Iraklis Psaroudakis, Florian Wolf, Norman May, Thomas Neumann, Alexander Böhm, Anastasia Ailamaki, and Kai-Uwe Sattler. 2014. Scaling Up Mixed Work- loads: A Battle of Data Freshness, Flexibility, and Scheduling. InPerformance Characterization and Benchmarking. Traditional to Big Data - 6th TPC Technology Conference, TPCTC 2014, Hangzhou, China, September...

    work page doi:10.1007/978-3-319-15350-6_7 2014

  48. [48]

    Ricardo Salazar-Díaz, Boris Glavic, and Tilmann Rabl. 2024. InferDB: In-Database Machine Learning Inference Using Indexes.Proc. VLDB Endow.17, 8 (2024), 1830–

    2024

  49. [49]

    doi:10.14778/3659437.3659441

    work page doi:10.14778/3659437.3659441

  50. [50]

    L. Sha, R. Rajkumar, and J.P. Lehoczky. 1990. Priority inheritance protocols: an approach to real-time synchronization.IEEE Trans. Comput.39, 9 (1990), 1175–1185. doi:10.1109/12.57058 Carl-Elliott Bilodeau-Savaria, Jan Kristof Nidzwetzki, Stefanie Scherzinger, and Bettina Kemme

    work page doi:10.1109/12.57058 1990

  51. [51]

    Haoze Song, Wenchao Zhou, Heming Cui, Xiang Peng, and Feifei Li. 2024. A survey on hybrid transactional and analytical processing.VLDB J.33, 5 (2024), 1485–1515. doi:10.1007/S00778-024-00858-9

    work page doi:10.1007/s00778-024-00858-9 2024

  52. [52]

    1995.Earliest Eligible Virtual Dead- line First: A Flexible and Accurate Mechanism for Proportional Share Resource Allocation

    Ion Stoica and Hussein Abdel-Wahab. 1995.Earliest Eligible Virtual Dead- line First: A Flexible and Accurate Mechanism for Proportional Share Resource Allocation. Technical Report TR-95-22. Old Dominion University. https: //people.eecs.berkeley.edu/~istoica/papers/eevdf-tr-95.pdf Revised January 1996

    1995

  53. [53]

    The Linux kernel community. [n. d.]. Control Groups Documentation. https: //docs.kernel.org/admin-guide/cgroup-v1/cgroups.html. Accessed: 2026-01-11

    2026

  54. [54]

    The Linux Kernel Development Community. 2025. Real-Time group scheduling. https://docs.kernel.org/scheduler/sched-rt-group.html. Accessed: 2026-04-21

    2025

  55. [55]

    Ken Thompson and Dennis M. Ritchie. 1973.UNIX Programmer’s Manual, Fourth Edition. Bell Telephone Laboratories, Inc., Murray Hill, New Jersey. https: //dspinellis.github.io/unix-v4man/v4man.pdf

    1973

  56. [56]

    Stephen Tu, Wenting Zheng, Eddie Kohler, Barbara Liskov, and Samuel Madden

  57. [57]

    InACM SIGOPS 24th Symposium on Operating Systems Principles, SOSP ’13, Farmington, PA, USA, November 3-6, 2013, Michael Kaminsky and Mike Dahlin (Eds.)

    Speedy transactions in multicore in-memory databases. InACM SIGOPS 24th Symposium on Operating Systems Principles, SOSP ’13, Farmington, PA, USA, November 3-6, 2013, Michael Kaminsky and Mike Dahlin (Eds.). ACM, 18–32. doi:10.1145/2517349.2522713

    work page doi:10.1145/2517349.2522713 2013

  58. [58]

    Benjamin Wagner, André Kohn, and Thomas Neumann. 2021. Self-Tuning Query Scheduling for Analytical Workloads. InSIGMOD ’21: International Conference on Management of Data, Virtual Event, China, June 20-25, 2021, Guoliang Li, Zhanhuai Li, Stratos Idreos, and Divesh Srivastava (Eds.). ACM, 1879–1891. doi:10.1145/3448016.3457260

    work page doi:10.1145/3448016.3457260 2021

  59. [59]

    Ziniu Wu, Markos Markakis, Chunwei Liu, Peter Baile Chen, Balakrishnan Narayanaswamy, Tim Kraska, and Samuel Madden. 2025. Improving DBMS Scheduling Decisions with Accurate Performance Prediction on Concurrent Queries.Proc. VLDB Endow.18, 11 (July 2025), 4185–4198. doi:10.14778/3749646. 3749686

    work page doi:10.14778/3749646 2025

  60. [60]

    Chenhao Ye, Wuh-Chwen Hwang, Keren Chen, and Xiangyao Yu. 2023. Polaris: Enabling Transaction Priority in Optimistic Concurrency Control.Proc. ACM Manag. Data1, 1 (2023), 44:1–44:24. doi:10.1145/3588724

    work page doi:10.1145/3588724 2023

  61. [61]

    Chen Yu. 2023. [PATCH v8 0/2] sched/fair: Introduce SIS_CURRENT to wake up short task on current CPU. https://lkml.org/lkml/2023/4/28/518. LKML posting (v8), with links back to earlier versions including v1 (Sep 2022)

    2023

  62. [62]

    Chen Yu. 2023. Re: [RFC PATCH] sched/fair: Bias runqueue selection towards almost idle prev CPU. https://lists.openwall.net/linux-kernel/2023/09/30/89

    2023

  63. [63]

    Yu, Kun-Lung Wu, Kwei-Jay Lin, and Sang H

    Philip S. Yu, Kun-Lung Wu, Kwei-Jay Lin, and Sang H. Son. 1994. On real-time databases: concurrency control and scheduling.Proc. IEEE82, 1 (1994), 140–157. doi:10.1109/5.259432

    work page doi:10.1109/5.259432 1994

  64. [64]

    Yunjia Zheng, Yuxuan Tian, Joseph Vinish D’silva, and Bettina Kemme. 2023. DBMLSched: Scheduling In-database Machine Learning Jobs. InJoint Proceed- ings of Workshops at the 49th International Conference on Very Large Data Bases (VLDB 2023), Vancouver, Canada, August 28 - September 1, 2023 (CEUR Workshop Proceedings). CEUR-WS.org. https://ceur-ws.org/Vol-...

    2023

  65. [65]

    Xinjing Zhou, Viktor Leis, Jinming Hu, Xiangyao Yu, and Michael Stonebraker

  66. [66]

    ACM Manag

    Practical DB-OS Co-Design with Privileged Kernel Bypass.Proc. ACM Manag. Data3, 1 (2025), 64:1–64:27. doi:10.1145/3709714

    work page doi:10.1145/3709714 2025

  67. [67]

    Peter Zijlstra. 2016. Re: [RFC] The Linux Scheduler: a Decade of Wasted Cores Report. Linux Kernel Mailing List (LKML). https://lkml.org/lkml/2016/4/25/176 Email dated 2016-04-25

    2016