Home > Research > Publications & Outputs > Wayfinder

Associated organisational unit

Electronic data

  • apsys21-final10

    Rights statement: Permission to make digital or hard copies of part or all of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for third-party components of this work must be honored. For all other uses, contact the owner/author(s).

    Accepted author manuscript, 1.02 MB, PDF document

    Available under license: CC BY-NC-SA: Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License

Links

Text available via DOI:

View graph of relations

Wayfinder: Towards Automatically Deriving Optimal OS Configurations

Research output: Contribution in Book/Report/Proceedings - With ISBN/ISSNConference contribution/Paperpeer-review

Published
Close
NullPointerException

Abstract

Tuning operating systems configuration in order to obtain the maximum application performance is a hard problem. This is due to the extremely large size of the configuration space offered by modern OSes, and to the fact that it is generally explored manually. To address that issue, we propose to bring automation to the OS configuration space exploration process, in order to derive effortlessly and as quickly as possible optimal OS configurations for a given use case.

We present Wayfinder, a generic OS performance evaluation platform. Wayfinder is fully automated and ensures both the accuracy and reproducibility of results, all the while speeding up how fast tests are run on a system. Wayfinder is easily extensible and offers convenient APIs to (1) implement custom configuration space exploration techniques, (2) add new benchmarks and (3) support additional OS projects. We demonstrate Wayfinder’s capacity to automatically and efficiently explore a LibOS’ networking configuration space; as well as its ability to efficiently isolate parallel experiments to avoid noisy neighbors.