When a Borrowed Slice Splits Replicas: Deterministic Resource Metering in a Replicated State Machine

In replicated state machines, resource accounting is part of the replicated protocol: a command's admission or refusal must be deterministic across replicas, regardless of hardware, memory residency, or implementation details. This paper reports a latent consensus defect in a production Rust-based system using Raft, where a storage read was charged differently depending on whether its payload was borrowed from memory or copied after eviction. We present a deterministic resource metering design based on portable integer prices over logical operations, charged before their effects, with admission rechecked during replicated apply before state mutation. Resource prices are treated as versioned protocol data, protected by known-answer tests and compatibility rules. Physical memory usage is validated separately against an empirical envelope rather than used to determine replicated outcomes. Experimental results demonstrate the elimination of residency-dependent admission differences, substantial reductions in memory allocated before refusal, and consistent accounting across tested execution targets. We discuss the assurance levels, limitations, and remaining engineering challenges of this approach.

Authors

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-10-14
DOI
https://doi.org/10.5281/zenodo.23324282
Primary Topic
Distributed systems and fault tolerance
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
preprint

When a Borrowed Slice Splits Replicas: Deterministic Resource Metering in a Replicated State Machine

Sebastian Thiebaud
Zenodo (CERN European Organization for Nuclear Research)
Distributed systems and fault tolerance
preprint

When a Borrowed Slice Splits Replicas: Deterministic Resource Metering in a Replicated State Machine

Sebastian Thiebaud
preprint en

Abstract

In replicated state machines, resource accounting is part of the replicated protocol: a command's admission or refusal must be deterministic across replicas, regardless of hardware, memory residency, or implementation details. This paper reports a latent consensus defect in a production Rust-based system using Raft, where a storage read was charged differently depending on whether its payload was borrowed from memory or copied after eviction. We present a deterministic resource metering design based on portable integer prices over logical operations, charged before their effects, with admission rechecked during replicated apply before state mutation. Resource prices are treated as versioned protocol data, protected by known-answer tests and compatibility rules. Physical memory usage is validated separately against an empirical envelope rather than used to determine replicated outcomes. Experimental results demonstrate the elimination of residency-dependent admission differences, substantial reductions in memory allocated before refusal, and consistent accounting across tested execution targets. We discuss the assurance levels, limitations, and remaining engineering challenges of this approach.

Zenodo (CERN European Organization for Nuclear Research)
Distributed systems and fault tolerance
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.