Understanding TCP incast throughput collapse in datacenter networks

TLDR

TCP Incast causes severe link under‑utilization in many‑to‑one traffic patterns and has been observed in distributed storage, MapReduce, and web‑search workloads. The study aims to understand Incast dynamics and develop an analytical model to explain symptoms, identify contributing factors, and evaluate proposed solutions. The authors analyze empirical data of concurrent TCP flows and construct an analytical model to capture Incast behavior, identify contributing factors, and assess solution effectiveness.

Abstract

TCP Throughput Collapse, also known as Incast, is a pathological behavior of TCP that results in gross under-utilization of link capacity in certain many-to-one communication patterns. This phenomenon has been observed by others in distributed storage, MapReduce and web-search workloads. In this paper we focus on understanding the dynamics of Incast. We use empirical data to reason about the dynamic system of simultaneously communicating TCP entities. We propose an analytical model to account for the observed Incast symptoms, identify contributory factors, and explore the efficacy of solutions proposed by us and by others.