Large MTUs and internet performance

@article{Murray2012LargeMA,
  title={Large MTUs and internet performance},
  author={David Murray and Terry Koziniec and Kevin Lee and Michael Dixon},
  journal={2012 IEEE 13th International Conference on High Performance Switching and Routing},
  year={2012},
  pages={82-87}
}
  • D. Murray, T. Koziniec, +1 author M. Dixon
  • Published 2012
  • Computer Science
  • 2012 IEEE 13th International Conference on High Performance Switching and Routing
Ethernet data rates have increased many orders of magnitudes since standardisation in 1982. Despite these continual data rates increases, the 1500 byte Maximum Transmission Unit (MTU) of Ethernet remains unchanged. Experiments with varying latencies, loss rates and transaction lengths are performed to investigate the potential benefits of Jumboframes on the Internet. This study reveals that large MTUs offer throughputs much larger than a simplistic overhead analysis might suggest. The reasons… Expand
Performance of Jumbo Sized Data on Jumbo Frame and Ethernet Frame Using UDP over IPv4/IPv6
TLDR
This study reveals that Jumbo frame has low delay and jitter values than Ethernet frame for some packet sizes and is tested on IPv4 and IPv6 network using UDP protocol. Expand
Performance comparison of transmitting jumbo frame on Windows and Linux System
TLDR
The results prove that the OS used on implementation of jumbo frame affects on the network performance, and the implementation can be used to justify the impact ofJumbo frame on thenetwork as well as operating systems performance. Expand
Impact of network protocols on data center applications
TLDR
Evaluating the advantages of jumbo frames using modern hardware with features such as large send/receive offload, and with canonical data center applications such as MapReduce and tiered Web services, finds that the throughput and CPU utilization benefits still exist generally, although compared to prior studies, are significantly reduced. Expand
Jumbo Frames or Not: That is the Question!
TLDR
It is concluded that data center network operators can safely turn jumbo frames on, despite a small side effect the authors discovered, and the throughput and CPU utilization benefits still exist generally, although compared to prior studies, are significantly reduced. Expand
Design of a Scalable Network Interface to Support Enhanced TCP and UDP Processing for High Speed Networks
TLDR
Two software approaches are presented to enhance protocol processing of TCP and UDP in the network interface and acceleration of the processing time frame required at the NI has been implemented to prevent hazards during the execution of the LRO and the LSO functions. Expand
The state of enterprise network traffic in 2012
  • D. Murray, T. Koziniec
  • Computer Science
  • 2012 18th Asia-Pacific Conference on Communications (APCC)
  • 2012
TLDR
This study has captured anonymised, and analysed Internet traffic entering and exiting a university network and a range of Network and Transport layer statistics, such as DCSP, SACK, ECN, MSS usage, Window Scaling and Timestamps are investigated. Expand
Curtailing latency in data center network by adopting Jumbo Frames
  • K. Sharma, V. Badarla
  • Computer Science
  • 2016 IEEE International Conference on Advanced Networks and Telecommunications Systems (ANTS)
  • 2016
TLDR
The proposed idea in this paper, JFEPM (Jumbo Frame Enabled Packet Merge), allows in-network data aggregation and with the help of jumbo frames it utilizes topology architecture of data center. Expand
Improving Network Efficiency with Simplemux
TLDR
Simplemux is presented, a protocol able to multiplex a number of packets sharing a common network path, thus increasing efficiency when small packets are transported, and can be useful in constrained scenarios where resources are scarce, as community wireless networks or IoT. Expand
Block Level Streaming Based Alternative Approach for Serving a Large Number of Workstations Securely and Uniformly
TLDR
The use of block streaming, normally used for storage, is presented to serve operating system and applications on-demand over the network to a workstation, also referred to as a client, a client computer, or a client workstation. Expand
Solving MTU Mismatch and Broadcast Overhead of NDN over Link-layer Networks
TLDR
Named Data Networking (NDN) is one of the future internet architectures using the data centric network model and can be implemented on top of the current Transmission Control Protocol/ Internet Protocol (TCP/IP) stack, or deployed directly over a linklayer network without the TCP/IP protocol stack. Expand
...
1
2
3
4
...

References

SHOWING 1-10 OF 30 REFERENCES
Increasing TCP's Initial Window
TLDR
This document specifies an optional standard for TCP to increase the permitted initial window from one or two segment(s) to roughly 4K bytes, replacing RFC 2414, and provides guidance on implementation issues. Expand
TCP performance re-visited
TLDR
It is concluded that unless major architectural changes are adopted, the Linux-2.4 TCP stack would be hard-pressed to continue relying on the 1 GHz/1 Gbps rule of thumb. Expand
A practical approach to TCP high speed WAN data transfers
— In this paper we describe the current state of the art in equipment, software, networks design and methods for transferring large scientific datasets at high speed around the globe, identifying theExpand
Packetization Layer Path MTU Discovery
TLDR
This method is described as an extension to RFC 1191 and RFC 1981, which specify ICMP-based Path MTU Discovery for IP versions 4 and 6, respectively. Expand
TCP Congestion Control
TLDR
This document defines TCP's four intertwined congestion control algorithms: slow start, congestion avoidance, fast retransmit, and fast recovery, as well as discussing various acknowledgment generation methods. Expand
Optimizing 10-Gigabit Ethernet for Networks of Workstations, Clusters, and Grids: A Case Study
TLDR
10GbE may be a cost-effective solution across a multitude of computing environments and is demonstrated that the 10GbE adapter can perform well in local-area, storage- area, system-area and wide-area networks. Expand
16 000-64 000 B pMTU experiments with simulation: The case for super jumbo frames at Supercomputing '05
This paper presents a case study including the results and preliminary simulations for a series of Ethernet-based Xnet ''super jumbo frame'' (SJF) experiments conducted prior to and at SupercomputingExpand
Performance of 1 and 10 Gigabit Ethernet cards with server quality motherboards
TLDR
This paper shows the importance of having a good combination of memory and peripheral bus chipset, Ethernet Interface, CPU power, good driver, and operating system designs and proper configuration to achieve and sustain gigabit transfer rates. Expand
The Ethernet Frame Payload Size and Its Effect on IPv4 and IPv6 Traffic
TLDR
Conclusions point to the possible existence of increased routing and switching effort resulting from the shift to the IPv6 protocol and suggest that a wider limit could decrease significantly the number of generated packets. Expand
Architectural characterization of TCP/IP packet processing on the Pentium/spl reg/ M microprocessor
  • S. Makineni, R. Iyer
  • Computer Science
  • 10th International Symposium on High Performance Computer Architecture (HPCA'04)
  • 2004
TLDR
An in-depth analysis of packet processing performance on Intel's state-of-the-art low power Pentium/spl reg/ M microprocessor running the Microsoft Windows* Server 2003 operating system finds that the mode of TCP/IP operation can significantly affect the performance requirements. Expand
...
1
2
3
...