ICTCP: Incast Congestion Control for TCP

TCP incast congestion happens in high-bandwidth and low-latency networks, when multiple synchronized servers send data to a same receiver in parallel \cite{incast09}. For many important data center applications such as MapReduce\cite{mapreduce04} and Search, this many-to-one traffic pattern is common. Hence TCP incast congestion may severely degrade their performances, e.g., by increasing response time.

In this paper, we study TCP incast in detail by focusing on the relationship among TCP throughput, round trip time (RTT) and receive window. Different from the previous approach to mitigate the impact of incast congestion by a fine grained timeout value, our idea is to design an ICTCP (Incast congestion Control for TCP) scheme at the receiver side. In particular, our method adjusts TCP receive window proactively before packet drops occur. The implementation and experiments in our testbed demonstrate that we achieve almost zero timeout and high goodput for TCP incast.

ictcp.pdf
PDF file

In  ACM CONEXT 2010

Publisher  Association for Computing Machinery, Inc.

Details

TypeInproceedings
> Publications > ICTCP: Incast Congestion Control for TCP