Fibre Channel Over Ethernet Research Articles

FCoE and iSCSI Performance Analysis in Tape Virtualization Systems

Tape virtualization systems' I/O performance have become almost the same as those observed in disk subsystems. However, few studies can be found in the literature on iSCSI performance with tape virtualization subsystems. Fibre Channel over Ethernet (FCoE) has been proposed recently as an alternative to both iSCSI and FC, but it remains challenging to find FCoE performance studies. Considering this gap in the literature, we did a comparative performance study between FCoE and iSCSI by executing I/O operations in virtual tapes using different block sizes. The goal was to determine the I/O average throughput, protocol overhead, I/O latency variance and relation between average throughput and CPU usage. The results showed an equivalent I/O throughput between FCoE and iSCSI operations, but FCoE surpassed iSCSI considering its lower overhead, processing and I/O latency variance values. It was also noted that FCoE software implementation must be enhanced.

Buffer Design Under Bursty Traffic with Applications in FCoE Storage Area Networks

This article studies the buffer requirements to design the Enhanced Ethernet required by Data Center applications, with bounded buffer overflow probability. The design criteria considers a bursty traffic arrival process, as it shows from conventional read/write SCSI operations in a Fibre Channel over Ethernet (FCoE) scenario. The results show that the buffer must be between one and two orders of magnitude larger than the maximum burst size to provide negligible buffer overflow probability. We complete our study by testing our assumptions with real FCoE measurement traces.

Survey on converged data center networks with DCB and FCoE: standards and protocols

Data center networks today face exciting new challenges in supporting cloud computing and other data-intensive applications. In conventional DCNs, different types of traffic are carried by different types of networks, such as Ethernet and Fibre Channel. Typically, Ethernet carries data traffic among servers in LANs, and Fibre Channel connects servers and storages in storage area networks. Due to the existence of multiple networks, the network cost, power consumption, wiring complexity, and management overhead are often high. The concept of a converged DCN is therefore appealing, carrying both types of traffic in a single converged Ethernet. Recent standards have been proposed for unified data center bridging (DCB) Ethernet and Fibre Channel over Ethernet (FCoE) protocols by the DCB Task Group of IEEE and the T11 Technical Committee of INCITS. In this article, we give a survey of the standards and protocols on converged DCNs, focusing mainly on their motivations and key functionalities. The technologies are discussed mainly from a practical perspective and may serve as a foundation for future research in this area.

Explicit coordination to prevent congestion in data center networks

Large cluster-based cloud computing platforms increasingly use commodity Ethernet technologies, such as Gigabit Ethernet, 10GigE, and Fibre Channel over Ethernet (FCoE), for intra-cluster communication. Traffic congestion can become a performance concern in the Ethernet due to consolidation of data, storage, and control traffic over a common layer-2 fabric, as well as consolidation of multiple virtual machines (VMs) over less physical hardware. Even as networking vendors race to develop switch-level hardware support for congestion management, we make the case that virtualization has opened up a complementary set of opportunities to reduce or even eliminate network congestion in cloud computing clusters. We present the design, implementation, and evaluation of a system called XCo, that performs explicit coordination of network transmissions over a shared Ethernet fabric to proactively prevent network congestion. XCo is a software-only distributed solution executing only in the end-nodes. A central controller uses explicit permissions to temporally separate (at millisecond granularity) the transmissions from competing senders through congested links. XCo is fully transparent to applications, presently deployable, and independent of any switch-level hardware support. We present a detailed evaluation of our XCo prototype across a number of network congestion scenarios, and demonstrate that XCo significantly improves network performance during periods of congestion. We also evaluate the behavior of XCo for large topologies using NS3 simulations.