Enabling configuration self-adaptation using machine learning
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Due to advancements in distributed systems and the increasing industrial demands placed on these systems, distributed systems are comprised of multiple complex components (e.g databases and their replication infrastructure, caching components, proxies, and load balancers) each of which have their own complex configuration parameters that enable them to be tuned for given runtime requirements. Software Engineers must manually tinker with many of these configuration parameters that change the behaviour and/or structure of the system in order to achieve their system requirements. In many cases, static configuration settings might not meet certain demands in a given context and ad hoc modifications of these configuration parameters can trigger unexpected behaviours, which can have negative effects on the quality of the overall system. In this work, I show the design and analysis of Finch; a tool that injects a machine learning based MAPE-K feedback loop to existing systems to automate how these configuration parameters are set. Finch configures and optimizes the system to meet service-level agreements in uncertain workloads and usage patterns. Rather than changing the core infrastructure of a system to fit the feedback loop, Finch asks the user to perform a small set of actions: instrumenting the code and configuration parameters, defining service-level objectives and agreements, and enabling programmatic changes to these configurations. As a result, Finch learns how to dynamically configure the system at runtime to self-adapt to its dynamic workloads. I show how Finch can replace the trial-and-error engineering effort that otherwise would be spent manually optimizing a system's wide array of configuration parameters with an automated self-adaptive system.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it