Attaining the Promise and Avoiding the Pitfalls of TCP in the Datacenter

Over the last several years, datacenter computing has become a pervasive part of the computing landscape. In spite of the success of the datacenter computing paradigm, there are significant challenges remaining to be solved--particularly in the area of networking. The success of TCP/IP in the Internet makes TCP/IP a natural candidate for datacenter network communication. A growing body of research and operational experience, however, has found that TCP often performs poorly in datacenter settings. TCP's poor performance has led some groups to abandon TCP entirely in the datacenter. This is not desirable, however, as it requires reconstruction of a new transport protocol as well as rewriting applications to use the new protocol. Over the last few years, promising research has focused on adapting TCP to operate in the datacenter environment. We have been running large datacenter computations for several years, and have experienced the promises and the pitfalls that datacenter computation presents. In this paper, we discuss our experiences with network communication performance within our datacenter, and discuss how we have leveraged and extended recent research to significantly improve network performance within our datacenter.

[1]  Albert G. Greenberg,et al.  Data center TCP (DCTCP) , 2010, SIGCOMM '10.

[2]  Nick McKeown,et al.  pFabric: minimal near-optimal datacenter transport , 2013, SIGCOMM.

[3]  Aleksandar Kuzmanovic,et al.  Adding Explicit Congestion Notification (ECN) Capability to TCP's SYN/ACK Packets , 2005, RFC.

[4]  Navendu Jain,et al.  Understanding network failures in data centers , 2011, SIGCOMM 2011.

[5]  Aleksandar Kuzmanovic,et al.  The power of explicit congestion notification , 2005, SIGCOMM '05.

[6]  David L. Black,et al.  The Addition of Explicit Congestion Notification (ECN) to IP , 2001, RFC.

[7]  GillPhillipa,et al.  Understanding network failures in data centers , 2011 .

[8]  Injong Rhee,et al.  CUBIC: a new TCP-friendly high-speed TCP variant , 2008, OPSR.

[9]  Navendu Jain,et al.  Understanding network failures in data centers: measurement, analysis, and implications , 2011, SIGCOMM.

[10]  Srinivasan Seshan,et al.  Measurement and Analysis of TCP Throughput Collapse in Cluster-based Storage Systems , 2008, FAST.

[11]  Adel Javanmard,et al.  Analysis of DCTCP: stability, convergence, and fairness , 2011, PERV.

[12]  Minlan Yu,et al.  Profiling Network Performance for Multi-tier Data Center Applications , 2011, NSDI.

[13]  D. Zats,et al.  DeTail: reducing the flow completion time tail in datacenter networks , 2012, CCRV.

[14]  Tao Yang,et al.  The Panasas ActiveScale Storage Cluster - Delivering Scalable High Bandwidth Storage , 2004, Proceedings of the ACM/IEEE SC2004 Conference.

[15]  Luiz André Barroso,et al.  The Datacenter as a Computer: An Introduction to the Design of Warehouse-Scale Machines , 2009, The Datacenter as a Computer: An Introduction to the Design of Warehouse-Scale Machines.

[16]  Matthew Mathis,et al.  Automatic TCP buffer tuning , 1998, SIGCOMM '98.

[17]  Douglas J. Leith,et al.  Experimental evaluation of Cubic-TCP , 2008 .

[18]  Antony I. T. Rowstron,et al.  Better never than late: meeting deadlines in datacenter networks , 2011, SIGCOMM.

[19]  Brighten Godfrey,et al.  Finishing flows quickly with preemptive scheduling , 2012, CCRV.

[20]  Haitao Wu,et al.  Tuning ECN for data center networks , 2012, CoNEXT '12.

[21]  Amar Phanishayee,et al.  Safe and effective fine-grained TCP retransmissions for datacenter communication , 2009, SIGCOMM '09.