Experimental and Theoretical Analysis of Storage Friendly TCP Performance in Distributed Storage Area Network

Fibre channel storage area networks (SAN) are widely implemented in production data center environments. Recently the storage industry has moved towards deployment of distributed SANs (DSAN), geographically dispersed across large physical distances. In a DSAN, specialized gateway devices interconnect the individual Fibre Channel (FC) fabrics over IP networks using TCP/IP based protocols (iFCP or FCIP) or over metro to long distance optical networks such as Dense Wavelength Division Multiplexing (DWDM) based networks that utilize native FC ports supporting large numbers of link credits. When using TCP/IP based storage networking protocols to interconnect local FC fabrics in a DSAN, the sustained throughput achievable depends upon the link characteristics and TCP/IP stack implementation. Sustaining maximum possible storage traffic throughput across the wide area network enables practical DSAN deployments by maintaining the required site to site service level agreements. This study explores the effects of several TCP/IP modifications on sustained traffic throughput for a DSAN interconnected via iFCP gateways across an impaired network. The TCP/IP stack modifications, known as storage friendly, include changes to the window scaling, congestion avoidance, and fast recovery algorithms. The theoretical background and experimental results are presented to explain and illustrate these modifications.

[1]  Tom Aug. Clark.  IP SANs : a guide to iSCSI, iFCP, and FCIP protocols for storage area networks , 2002 .

[2]  Yoohwan Kim,et al.  TCP performance comparison under various load balancing methods using OPNET , 2002 .

[3]  R. Srikant,et al.  Multi-Path TCP: A Joint Congestion Control and Routing Scheme to Exploit Path Diversity in the Internet , 2006, IEEE/ACM Transactions on Networking.

[4]  Jing Zhu,et al.  Performance modelling of TCP enhancements in terrestrial-satellite hybrid networks , 2006, TNET.

[5]  Suresh Muknahallipatna,et al.  The effect of end to end latency in a distributed storage area network on Microsoft Exchange Server 2003 performance - Part I , 2004, 29th Annual IEEE International Conference on Local Computer Networks.

[6]  T. V. Lakshman,et al.  The performance of TCP/IP for networks with high bandwidth-delay products and random loss , 1997, TNET.

[7]  K. K. Ramakrishnan,et al.  Explicit window adaptation: a method to enhance TCP performance , 1998, Proceedings. IEEE INFOCOM '98, the Conference on Computer Communications. Seventeenth Annual Joint Conference of the IEEE Computer and Communications Societies. Gateway to the 21st Century (Cat. No.98.

[8]  J. Abella,et al.  Simulation-based study of TCP flow control mechanisms using OPNET Modeler , 2002 .

[9]  Victor O. K. Li,et al.  An Overview of Packet Reordering in Transmission Control Protocol (TCP): Problems, Solutions, and Challenges , 2007, IEEE Transactions on Parallel and Distributed Systems.

[10]  Sung-Soo Kim,et al.  Congestion Avoidance Algorithm Using Extended Kalman Filter , 2007, 2007 International Conference on Convergence Information Technology (ICCIT 2007).

[11]  Cheng Jin,et al.  FAST TCP: Motivation, Architecture, Algorithms, Performance , 2006, IEEE/ACM Transactions on Networking.

[12]  Robert W. Kembel.  Fibre Channel Switched Fabric , 2001 .