Tese de mestrado. Redes e Serviços de Comunicação. Faculdade de Engenharia. Universidade do Porto, INESC Porto. 2009

The IEEE Standard for Gigabit Networking was accepted in June 1998 and ratified as IEEE 802.3z. This standard uses considers an optical cable for signal transmission. One year later a new standard for Gigabit Ethernet over unshielded twisted pair of the 5th category was certified under the name 802.3ab. Nowadays, there are a couple of younger and older studies about Gigabit Ethernet technology and performance. This Meta Analysis is intended to put the main results altogether into one document suitable for a proper choice of gigabit networking equipment for cluster computers.

This work covers the implementation of a raw Ethernet communication module for the Open MPI message passing library. Thereby it focuses on both the reduction of the communication latency for small messages and maximum possible compatibility. Especially the need for particular network devices, adapted network device drivers or kernel patches is avoided. The work is divided into three major parts: First, the networking subsystem of the version 2.6 Linux kernel is analyzed. Second, an Ethernet protocol family is implemented as a loadable kernel module, consisting of a basic datagram protocol (EDP), providing connection-less and unreliable datagram transport, and a streaming protocol (ESP), providing connection-oriented, sequenced and reliable byte streams. The protocols use the standard device driver interface of the Linux kernel for data transmission and reception. Their services are made available to user-space applications through the standard socket interface. Last, the existing Open MPI TCP communication module is ported atop the ESP. With bare EDP/ESP sockets a message latency of about 30 us could be achieved for small messages, which compared to the TCP latency of about 40 us is a reduction of 25 %.

The demand for ubiquitous networks has pushed the designs of networks all the way. The requirement for access point to be integrated into IEEE 802.3 standard and other networks has always been a sore point in the limitation of wireless coverage of IEEE 802.11 standard networks. Wireless Mesh Network (WMN) is expected to be the future of the next generation wireless network. It is experiencing a fast growing development due to its attractive features which includes high reliable connectivity, easy deployment, self healing, self configuring, flexible network expansion etc. Hence the mobility of the WMN nodes has been of paramount importance, which would make it independent of wired infrastructure and flexible interoperability with various networks and devices. The requirements like mobility, transparency etc. have led to the amendment of the WMN standard by the Institute of Electronics and Electrical Engineering (IEEE) 802.11 Working Group (WG), Task Group (TG) “S. The IEEE 802.11s standard tackles these issues by its operation on layer 2 of Open Systems Interconnection (OSI) reference model and creates a transparent IEEE 802 broadcast domain that supports any higher layer protocol. In our work we give the evaluation of the upcoming IEEE 802.11s standard based on its features some of which include routing at layer 2 and medium access control to enable its design and implementation in the existing mesh frame work of Communication Research Labs (CRL) using the proposed IEEE 802.11s routing protocols with focus on RA-OLSR and HWMP. We concentrate on how to integrate these features into the existing CRL’s C-CORE which runs other layer 3 routing protocols and complex functions as Application Programming Interface (API) modules. The implementation of the IEEE 802.11s standard creates major challenges as we have to create a roadmap on integrating the new wireless kernel interfaces like the nl80211, cfg80211 and the Wireless Extension (Wext) into the CRL’s C-CORE framework for communication between user space and kernel space, especially taking into consideration of the existing HAL and madwifi wireless drivers of the CRL’s framework. To support the evaluation of the features like the layer 2 routing and the modified MAC performance, we compare results of the CRL’s real time mesh network test with our simulation result of the IEEE 802.11s standard using the Qualnet 4.5 simulator with focus on the basic network parameters like delay, jitter and throughput. The comparison shows that the CRL’s network has higher throughput running its existing layer 3 protocols. The analysis also proves that the 802.11s is flexible, scalable and efficient in delivering multi hop capabilities to clients that cannot afford the deployment time or the cost for wired networks that use access points. With the complete integration, of the 802.11s standard specifications the CRL’s C-CORE framework can be much more capable of supporting more diverse network scenario deployments. / +46-736318897

The IEEE Standard for Gigabit Networking was accepted in June 1998 and ratified as IEEE 802.3z. This standard uses considers an optical cable for signal transmission. One year later a new standard for Gigabit Ethernet over unshielded twisted pair of the 5th category was certified under the name 802.3ab. Nowadays, there are a couple of younger and older studies about Gigabit Ethernet technology and performance. This Meta Analysis is intended to put the main results altogether into one document suitable for a proper choice of gigabit networking equipment for cluster computers.

This work covers the implementation of a raw Ethernet communication module for the Open MPI message passing library. Thereby it focuses on both the reduction of the communication latency for small messages and maximum possible compatibility. Especially the need for particular network devices, adapted network device drivers or kernel patches is avoided. The work is divided into three major parts: First, the networking subsystem of the version 2.6 Linux kernel is analyzed. Second, an Ethernet protocol family is implemented as a loadable kernel module, consisting of a basic datagram protocol (EDP), providing connection-less and unreliable datagram transport, and a streaming protocol (ESP), providing connection-oriented, sequenced and reliable byte streams. The protocols use the standard device driver interface of the Linux kernel for data transmission and reception. Their services are made available to user-space applications through the standard socket interface. Last, the existing Open MPI TCP communication module is ported atop the ESP. With bare EDP/ESP sockets a message latency of about 30 us could be achieved for small messages, which compared to the TCP latency of about 40 us is a reduction of 25 %.