Understanding TCP incast throughput collapse in datacenter networks
- 21 August 2009
- conference paper
- conference paper
- Published by Association for Computing Machinery (ACM)
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.Keywords
This publication has 6 references indexed in Scilit:
- Safe and effective fine-grained TCP retransmissions for datacenter communicationPublished by Association for Computing Machinery (ACM) ,2009
- A scalable, commodity data center network architecturePublished by Association for Computing Machinery (ACM) ,2008
- MapReduceCommunications of the ACM, 2008
- The Google file systemACM SIGOPS Operating Systems Review, 2003
- Analysis of the increase and decrease algorithms for congestion avoidance in computer networksComputer Networks and ISDN Systems, 1989
- Congestion avoidance and controlPublished by Association for Computing Machinery (ACM) ,1988