Cortex: Workflow-Aware Resource Pooling and Scheduling for Agentic Serving

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Pagonas, Nikos, Chung, Yeounoh, Kaffes, Kostis, Krishnamurthy, Arvind
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866917018484080640
author Pagonas, Nikos
Chung, Yeounoh
Kaffes, Kostis
Krishnamurthy, Arvind
author_facet Pagonas, Nikos
Chung, Yeounoh
Kaffes, Kostis
Krishnamurthy, Arvind
contents We introduce Cortex, a prototype workflow-aware serving platform designed for agentic workloads. The core principle of Cortex is stage isolation: it provisions dedicated resource pools for each distinct stage of an agentic workflow. This simple yet powerful strategy mitigates inter-stage interference in compute and memory, leading to better KV cache utilization, higher throughput, and more predictable performance. By customizing resource allocation and scheduling within each distinct stage of agentic workflows, Cortex lays the groundwork for more advanced, agent-native serving paradigms, including malleable resource management, speculative execution of workflow branches, and a shared, multi-tiered cache for "agentic state."
format Preprint
id arxiv_https___arxiv_org_abs_2510_14126
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Cortex: Workflow-Aware Resource Pooling and Scheduling for Agentic Serving
Pagonas, Nikos
Chung, Yeounoh
Kaffes, Kostis
Krishnamurthy, Arvind
Distributed, Parallel, and Cluster Computing
We introduce Cortex, a prototype workflow-aware serving platform designed for agentic workloads. The core principle of Cortex is stage isolation: it provisions dedicated resource pools for each distinct stage of an agentic workflow. This simple yet powerful strategy mitigates inter-stage interference in compute and memory, leading to better KV cache utilization, higher throughput, and more predictable performance. By customizing resource allocation and scheduling within each distinct stage of agentic workflows, Cortex lays the groundwork for more advanced, agent-native serving paradigms, including malleable resource management, speculative execution of workflow branches, and a shared, multi-tiered cache for "agentic state."
title Cortex: Workflow-Aware Resource Pooling and Scheduling for Agentic Serving
topic Distributed, Parallel, and Cluster Computing
url https://arxiv.org/abs/2510.14126