# SCION: Size-aware Policy Orchestration for Nonstationary Object Caches (Long Paper Version)

arXiv: https://arxiv.org/abs/2605.01055 · HTML: https://mlsanity.com/paper/2605.01055/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 8 of its 27 references that are in the map.
- First posted on arXiv 27 Mar 2026.

## Builds on (most cited first, 8 of 8)

- [An Imitation Learning Approach for Cache Replacement](https://mlsanity.com/paper/2006.16239/) (2020)
- [A Learned Cache Eviction Framework with Minimal Overhead](https://mlsanity.com/paper/2301.11886/) (2023)
- [A Learning-Based Caching Mechanism for Edge Content Delivery](https://mlsanity.com/paper/2402.02795/) (2024)
- [Toward Robust and Efficient ML-Based GPU Caching for Modern Inference](https://mlsanity.com/paper/2509.20979/) (2025)
- [Adaptive TTL-Based Caching for Content Delivery](https://mlsanity.com/paper/1704.04448/) (2017)
- [Cold-RL: Learning Cache Eviction with Offline Reinforcement Learning for NGINX](https://mlsanity.com/paper/2508.12485/) (2025)
- [DynamicAdaptiveClimb: Adaptive Cache Replacement with Dynamic Resizing](https://mlsanity.com/paper/2511.21235/) (2025)
- [Clock2Q+: A Simple and Efficient Replacement Algorithm for Metadata Cache in VMware vSAN](https://mlsanity.com/paper/2511.21958/) (2025)

## Abstract (arXiv)

Object caches underpin cloud and edge services, but production workloads are heterogeneous, nonstationary, and throughput-constrained. Recent simple non-ML policies such as SIEVE and S3-FIFO set a strong baseline, so any learned method must be overhead-aware, robust under drift, and competitive with strong experts. We present SCION, a lightweight policy-orchestration framework that selects among a small set of deployable cache policies using a tiny workload fingerprint computed off the critical path. Our prototype, AUTO, uses short-prefix statistics of object size, cacheability, reuse, and cache size, then applies an offline-trained linear selector to choose among GDSF, S3-FIFO, SIEVE, LHD, W-TinyLFU-AV, and DynamicAdaptiveClimb; a simpler SCION-P90 variant uses only a p90 threshold. In a CPU-only, trace-driven evaluation on 30 public object-cache traces and a separate HR-Cache simulator subset, AUTO improves cacheable-only object miss ratio over SIEVE on a majority of workloads, stays close to the best single expert on average, enables explicit OMR/BMR tradeoff selection, and remains competitive on byte miss ratio. Under a fast-policy budget, AUTO-fast achieves lower cost than the best fixed fast policy. SCION reduces regime-mismatch risk while keeping the hot path unchanged.