A Bring-Your-Own-Model Approach for ML-Driven Storage Placement in Warehouse-Scale Computers

Chenxi Yang, Yan Li, Martin Maas, Mustafa Uysal, Arif Merchant, Richard McDougall

Conference on Machine Learning and Systems 2025 · Day 3 · Session 6: Edge and Cloud Systems

In the sprawling landscape of modern data centers, where "warehouse-scale computers" are the norm, storage systems represent a significant portion of the total operational cost. Within these complex infrastructures, the decision of where to place data – specifically, whether to store it on fast, expensive Solid State Drives (SSDs) or slower, cheaper Hard Disk Drives (HDDs) – is a critical factor profoundly impacting both performance and cost. This talk, presented by Chenxi Yang and co-authored with Yan Li, Martin Maas, Mustafa Uysal, Arif Merchant, and Richard McDougall, introduces a novel **Bring Your Own Model (BYOM)** approach to machine learning (ML)-driven storage placement, aiming to optimize this crucial decision.

AI review

A competent systems ML paper presented as a conference talk — real engineering, real production results, but the write-up reads more like an abstract expansion than a talk review, and the engineering details that would let you actually build this are conspicuously absent. The core insight (decompose monolithic storage placement models into per-app GBTs that emit ranked hints, let an adaptive algorithm at the storage layer do the global coordination) is genuinely practical and the production numbers are credible. But this is Google infrastructure research, not something an engineer can pick…