Porting LibRIPC to iWARP

Cloud computing has become a major economical factor in the recent development of computer systems. Companies tend to draw computational power for their data processing from the cloud, instead of hosting their own servers, in order to save costs. The providers of cloud systems run huge and cost efficient data centers, profiting from economies of scale. In order to further reduce the costs, these data centers currently move to more power efficient systems. Many applications that are now used in the cloud were not created with a cloud environment in mind, but have been "moved" to the cloud. These applications usually use TCP/IP for their intercommunication mechanisms, which is the defacto standard for current applications in the Internet. Unfortunately, these TCP/IP based implementations rely on the Berkeley socket API, which does not match the demands of power efficient systems. Sockets introduce much CPU involvement, taking away precious computational time from the real applications. Several specialized network architectures, such as InfiniBand, overcome this issue. They make use of offloading techniques, such as RDMA, in which the network hardware takes over the responsibility of protocol processing and I/O. As a drawback, the equipment for these fabrics is usually expensive, due to their high specialization. Furthermore, the corresponding interfaces for these architectures are extensive, making it expensive to adapt existing applications. LibRIPC is a network library introduced to overcome the efforts that are accompanied by the usage of these architecture’s interfaces. It provides a neat, message based interface, which abstracts from any specifics of the underlying network architecture. This provides ease of integration and portability to other network fabrics, and yet does not sacrifice performance. Currently, there exist a prototype implementation of LibRIPC for InfiniBand. In this thesis, we present our port of LibRIPC to iWARP, which enables the library for its use over Ethernet. Ethernet is one of the most cost efficient network fabrics and provides capabilities for high performance networking since the upcoming of standards that specify data rates of 10 Gbit/s and above. iWARP is a protocol stack that is based upon TCP/IP, thus compatible for use in Internet based WANs. Yet it is completely independent from the socket API and its correspond-

[1]  Luiz André Barroso,et al.  The Datacenter as a Computer: An Introduction to the Design of Warehouse-Scale Machines, Second Edition , 2013, The Datacenter as a Computer: An Introduction to the Design of Warehouse-Scale Machines, Second Edition.

[2]  Randy H. Katz,et al.  Above the Clouds: A Berkeley View of Cloud Computing , 2009 .

[3]  Robert L. Grossman,et al.  The Case for Cloud Computing , 2009, IT Professional.

[4]  Pete Wyckoff,et al.  A Performance Analysis of the Ammasso RDMA Enabled Ethernet Adapter and its iWARP API , 2005, 2005 IEEE International Conference on Cluster Computing.

[5]  Odysseas I. Pentakalos An Introduction to the InfiniBand Architecture , 2002, Int. CMG Conference.

[6]  George L.-T. Chiu,et al.  Overview of the Blue Gene/L system architecture , 2005, IBM J. Res. Dev..

[7]  Alan Wagner,et al.  A Hybrid MPI design using SCTP and iWARP , 2008, 2008 IEEE International Symposium on Parallel and Distributed Processing.

[8]  Renato Recio,et al.  Marker PDU Aligned Framing for TCP Specification , 2007, RFC.

[9]  Lyndon Ong,et al.  An Introduction to the Stream Control Transmission Protocol (SCTP) , 2002, RFC.

[10]  Hemal Shah,et al.  Direct Data Placement over Reliable Transports , 2007, RFC.

[11]  Pete Wyckoff,et al.  Initial Performance Evaluation of the NetEffect 10 Gigabit iWARP Adapter , 2006, 2006 IEEE International Conference on Cluster Computing.

[12]  Craig Partridge,et al.  When the CRC and TCP checksum disagree , 2000, SIGCOMM.

[13]  Dhabaleswar K. Panda,et al.  Performance characterization of a 10-Gigabit Ethernet TOE , 2005, 13th Symposium on High Performance Interconnects (HOTI'05).

[14]  Brett D. Fleisch,et al.  The Chubby lock service for loosely-coupled distributed systems , 2006, OSDI '06.

[15]  Jon Postel,et al.  User Datagram Protocol , 1980, RFC.

[16]  Hyun-Wook Jin,et al.  Supporting iWARP Compatibility and Features for Regular Network Adapters , 2005, 2005 IEEE International Conference on Cluster Computing.

[17]  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.

[18]  Burkhard D. Steinmacher-Burow,et al.  The IBM Blue Gene/Q Interconnection Fabric , 2012, IEEE Micro.

[19]  Jeffrey C. Mogul,et al.  TCP Offload Is a Dumb Idea Whose Time Has Come , 2003, HotOS.

[20]  Alan Wagner,et al.  SCTP versus TCP for MPI , 2005, ACM/IEEE SC 2005 Conference (SC'05).

[21]  Pavan Balaji,et al.  iWARP redefined: Scalable connectionless communication over high-speed Ethernet , 2010, 2010 International Conference on High Performance Computing.

[22]  George Bosilca,et al.  High Performance RDMA Protocols in HPC , 2006, PVM/MPI.

[23]  Hakim Weatherspoon,et al.  NetSlices: Scalable multi-core packet processing in user-space , 2012, 2012 ACM/IEEE Symposium on Architectures for Networking and Communications Systems (ANCS).

[24]  Ahmad Afsahi,et al.  10-Gigabit iWARP Ethernet: Comparative Performance Analysis with InfiniBand and Myrinet-10G , 2007, 2007 IEEE International Parallel and Distributed Processing Symposium.

[25]  Philip Werner Frey,et al.  Zero-copy network communication: An applicability study of iWARP beyond micro benchmarks , 2010 .

[26]  Hyun-Wook Jin,et al.  Exploiting NIC architectural support for enhancing IP-based protocols on high-performance networks , 2005, J. Parallel Distributed Comput..

[27]  Dhabaleswar K. Panda,et al.  EMP: Zero-Copy OS-Bypass NIC-Driven Gigabit Ethernet Message Passing , 2001, ACM/IEEE SC 2001 Conference (SC'01).

[28]  Greg J. Regnier,et al.  TCP performance re-visited , 2003, 2003 IEEE International Symposium on Performance Analysis of Systems and Software. ISPASS 2003..

[29]  Pete Wyckoff,et al.  Design and Implementation of the iWarp Protocol in Software , 2005, IASTED PDCS.

[30]  Bernard Metzler,et al.  SoftRDMA: Implementing iWARP over TCP kernel sockets , 2010, IBM J. Res. Dev..

[31]  Brian Hayes,et al.  What Is Cloud Computing? , 2019, Cloud Technologies.

[32]  Alan D. George,et al.  Comparative Performance Analysis of RDMA-Enhanced Ethernet , 2005 .

[33]  David Clark,et al.  An analysis of TCP processing overhead , 1989 .

[34]  Ian Lumb,et al.  A Taxonomy and Survey of Cloud Computing Systems , 2009, 2009 Fifth International Joint Conference on INC, IMS and IDC.

[35]  Hakim Weatherspoon,et al.  Operating Systems Abstractions for Software Packet Processing in Datacenters , 2011 .

[36]  Pete Wyckoff,et al.  iWarp protocol kernel space software implementation , 2006, Proceedings 20th IEEE International Parallel & Distributed Processing Symposium.

[37]  David Hilley,et al.  Cloud Computing: A Taxonomy of Platform and Infrastructure-level Offerings , 2009 .

[38]  Brice Goglin Design and implementation of Open-MX: High-performance message passing over generic Ethernet hardware , 2008, 2008 IEEE International Symposium on Parallel and Distributed Processing.

[39]  Jeffrey C. Mogul,et al.  Remote Direct Memory Access (RDMA) over IP Problem Statement , 2005, RFC.

[40]  David D. Clark,et al.  An analysis of TCP processing overhead , 1988, IEEE Communications Magazine.

[41]  Frank Bellosa,et al.  Light-weight remote communication for high-performance cloud networks , 2012, 2012 IEEE 1st International Conference on Cloud Networking (CLOUDNET).

[42]  Arkady Kanevsky,et al.  Remote Direct Memory Access over the Converged Enhanced Ethernet Fabric: Evaluating the Options , 2009, 2009 17th IEEE Symposium on High Performance Interconnects.

[43]  Pavan Balaji,et al.  Sockets vs. RDMA Interface over 10-Gigabit Networks: An In-depth Analysis of the Memory Traffic Bottleneck , 2004 .

[44]  Lixia Zhang,et al.  Stream Control Transmission Protocol , 2000, RFC.

[45]  Greg J. Regnier,et al.  The Virtual Interface Architecture , 2002, IEEE Micro.

[46]  David D. Clark,et al.  The design philosophy of the DARPA internet protocols , 1988, SIGCOMM '88.