A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2022; you can also visit the original URL.
The file type is application/pdf
.
Filters
Enabling Programmable Transport Protocols in High-Speed NICs
2020
Symposium on Networked Systems Design and Implementation
In this paper, we enable programmable transport protocols in high-speed NICs by designing Tonic, a flexible hardware architecture for transport logic. ...
However, hardwiring the network stack in the NIC would stifle innovation in transport protocols. ...
We argue that transport protocols on high-speed NICs can be made programmable without exposing users to the full complexity of programming for high-speed hardware. ...
dblp:conf/nsdi/ArashlooLGRWW20
fatcat:f5342cdkm5a2bf3ueqaafefr2y
Long-haul secure data transfer using hardware-assisted GridFTP
2016
Future generations computer systems
Extreme-scale scientific collaborations require high-performance wide-area end-to-end data transports to enable fast and secure transfer of high data volumes among collaborating institutions. ...
*Manuscript Click here to view linked References the available bandwidth requires efficient underlying transport protocols and end-to-end implementations that can sustain a high bandwidth in high-latency ...
This work was supported in part by DOE Grants DE-FG02-05ER84163, DE-FG02-08ER86360, and DE-SC0002182. ...
doi:10.1016/j.future.2015.09.014
fatcat:kt23mfru75ahrjbhq4hi47nvja
TCP Offload Is a Dumb Idea Whose Time Has Come
2003
USENIX Workshop on Hot Topics in Operating Systems
In the context of the replacement of storage-specific interconnect via commoditized network hardware, TCP offload (and more generally, offloading the transport protocol) appropriately solves an important ...
But TCP offload in the service of very specific goals might actually be useful. ...
This argument is somewhat weakened by the reality that many (if not most) high-speed NICs are already reprogram- mable, but the extra capabilities of a TOE NIC might increase the options for subversion ...
dblp:conf/hotos/Mogul03
fatcat:zzedu5zaf5gibj6pvgn6jhxc4y
Integrated Management for Network Virtualization Infrastructure
2016
Journal of NICT
Our results achieved to meet five requirements which are resource abstraction, resource isolation, resource elasticity, deep programmability and authentication. ...
In addition, we proposed evolvable architecture and extended network virtualization for edge network and terminals. We also deployed our system on JGN-X testbed. ...
Figure 7 shows the structure of Programmer. Slow-Path provides a flexible programming environment which is realized through applying virtual machines (VM) and a high-speed network I/O environment. ...
doi:10.24812/nictjournal.62.2_033
fatcat:5q4hlg7n2bbfxnvelzkwtd7t4m
Design and Deployment of Enhanced VNode Infrastructure — Deeply Programmable Network Virtualization
2016
IEICE transactions on communications
We achieve coexistence of high performance and programmability. We also enhance AGW functions. In addition, we extend network virtualization from the core network to edge networks and terminals. ...
We present new extended VNode-i functions that offer high performance and provide convenient deep programmability to network developers. ...
In addition, high-speed packet forwarding is also required for the Programmer. ...
doi:10.1587/transcom.2015cci0002
fatcat:iynajs7u4rcjhiccrsu526p7jy
Impact of protocol overheads on network throughput over high-speed interconnects: measurement, analysis, and improvement
2007
Journal of Supercomputing
Such overheads are a big obstacle for high-performance computing applications to exploit high-speed interconnects in cluster environments. ...
In this paper, we exploit a set of protocol optimization mechanisms in an integrated manner while dealing with the full spectrum of the protocol layers from the transport layer to the physical layer. ...
In addition to the IP-based protocols, Myrinet also provides the high-speed user-level protocol called GM. ...
doi:10.1007/s11227-007-0101-0
fatcat:nhrk7aq4b5cylbj7aszqcuwlpy
A Linux Kernel Implementation of the Homa Transport Protocol
2021
USENIX Annual Technical Conference
Homa/Linux is a Linux kernel module that implements the Homa transport protocol. Measurements of Homa/Linux reconfirm Homa's superior performance compared to TCP and DCTCP. ...
Another factor of 5-10x in performance can be achieved if software overheads can be eliminated in the future. ...
In addition, existing NICs are not open or programmable. • Today's "smart NICs" have inadequate performance and/or programmability. Smart NICs come in two flavors. ...
dblp:conf/usenix/Ousterhout21
fatcat:jyhalgiojvcrzheyou6c7wiuvm
U-Net/SLE: A Java-Based User-Customizable Virtual Network Interface
1999
Scientific Programming
We describe U‐Net/SLE (Safe Language Extensions), a user‐level network interface architecture which enables per‐application customization of communication semantics through downloading of user extension ...
By leveraging the existing U‐Net model, applications may implement protocol code at the user level, within the NI, or using some combination of the two. ...
Introduction Recent work in high-speed interconnects for distributed and parallel computing systems, particularly workstation clusters, has focused on development of network interfaces enabling low-latency ...
doi:10.1155/1999/316853
fatcat:3h2d6k2snjerhbq7jlodp5556a
Exploiting task-level concurrency in a programmable network interface
2003
SIGPLAN notices
This paper aims to offset those performance disadvantages by exploiting task-level concurrency in the workload to parallelize the network interface firmware for a programmable controller with two processors ...
This parallelization results in performance within 10-20% of a modern ASIC-based network interface for real network services. ...
Queue Pair IP uses the existing Internet protocols as its transport layer, and the prototype implements components of TCP/IP directly in the LANai adapter. ...
doi:10.1145/966049.781506
fatcat:azlixmmn7rfctcg7sif74lpqti
Exploiting task-level concurrency in a programmable network interface
2003
Proceedings of the ninth ACM SIGPLAN symposium on Principles and practice of parallel programming - PPoPP '03
This paper aims to offset those performance disadvantages by exploiting task-level concurrency in the workload to parallelize the network interface firmware for a programmable controller with two processors ...
This parallelization results in performance within 10-20% of a modern ASIC-based network interface for real network services. ...
Queue Pair IP uses the existing Internet protocols as its transport layer, and the prototype implements components of TCP/IP directly in the LANai adapter. ...
doi:10.1145/781503.781506
fatcat:vtiht72jl5eadm3pcmmywggzae
Exploiting task-level concurrency in a programmable network interface
2003
Proceedings of the ninth ACM SIGPLAN symposium on Principles and practice of parallel programming - PPoPP '03
This paper aims to offset those performance disadvantages by exploiting task-level concurrency in the workload to parallelize the network interface firmware for a programmable controller with two processors ...
This parallelization results in performance within 10-20% of a modern ASIC-based network interface for real network services. ...
Queue Pair IP uses the existing Internet protocols as its transport layer, and the prototype implements components of TCP/IP directly in the LANai adapter. ...
doi:10.1145/781498.781506
dblp:conf/ppopp/KimPR03
fatcat:4dbjrzp37nfvplx5fe4jmppmcy
Programmable Network Data Planes (Dagstuhl Seminar 19141)
2019
Dagstuhl Reports
However, a key limitation of the original SDN vision was the limited functionality exposed in protocols such as OpenFlow. ...
In this seminar we discussed the key questions and problems to be addressed in the next 10 years on the area of programmable dataplanes, and how they will potentially shape the future of networking. ...
Some (Jon Crowcroft a while ago) involved writing a new transport protocol and interoperate. ...
doi:10.4230/dagrep.9.3.178
dblp:journals/dagstuhl-reports/AntichiBFRS19
fatcat:ulc3w7dvevduhowuymmf4wymzu
Towards 100 gbit/s ethernet
2009
Proceedings of the 23rd international conference on Conference on Supercomputing - ICS '09
While in principle suitable for high performance clustered and parallel applications, Ethernet requires matching improvements in the system software stack. ...
Key contributions of our work are the design of a parallel protocol that uses context-independent page-remapping to reduce packet processing overheads, thread management overhead and synchronization in ...
Previous research on programmable NICs [4, 16] showed that enabling application preposting of receive buffers directly with the NIC offers such a mechanism. ...
doi:10.1145/1542275.1542308
dblp:conf/ics/PassasMB09
fatcat:pi2kxxclwfcgjgp2223ejeae5u
Datacenter RPCs can be General and Fast
[article]
2019
arXiv
pre-print
It is commonly believed that datacenter networking software must sacrifice generality to attain high performance. ...
on commodity CPUs in traditional datacenter networks based on either lossy Ethernet or lossless fabrics. eRPC performs well in three key metrics: message rate for small messages; bandwidth for large messages ...
This solution does not work for the NIC DMA queue: since we use unsignaled transmission, it is generally impossible for software to know whether a request is in the DMA queue without ushing it. ...
arXiv:1806.00680v2
fatcat:ryqqflbh4rgqpf2voguhgdhs6e
Multistage Switching Architectures for Software Routers
2007
IEEE Network
We propose a specific multistage architecture, exploiting PC-based routers as switching elements, to build a high-speed, largesize, scalable, and reliable software router. ...
in terms of number of network interface cards, and lack of resilience mechanisms. ...
Acknowledgments This work was performed in the framework of the Bora-Bora project funded by the Italian Ministry of University, Education, and Research (MIUR) and developed in the high-quality lab LIPAR ...
doi:10.1109/mnet.2007.386465
fatcat:7lz7jla6jfdh7bhdwlj5olfxp4
« Previous
Showing results 1 — 15 out of 1,704 results