On the Benefit of Caching Traffic Flow Data in the Link Buffer

Authors:
Konstantin Septinus;Christian Grimm;Vladislav Rumyantsev;Peter Pirsch
Affiliations:
Institute of Microelectronic Systems, , Hannover, Germany 30167;Regional Computing Centre for Lower Saxony, Hannover, Germany 30159;Institute of Microelectronic Systems, , Hannover, Germany 30167;Institute of Microelectronic Systems, , Hannover, Germany 30167
Venue:
SAMOS '08 Proceedings of the 8th international workshop on Embedded Computer Systems: Architectures, Modeling, and Simulation
Year:
2008

Citing 6
Cited 0

TCP Switching: Exposing Circuits to IP

IEEE Micro
Cost-Effective Flow Table Designs for High-Speed Routers: Architecture and Performance Evaluation

IEEE Transactions on Computers
On the elusive benefits of protocol offload

NICELI '03 Proceedings of the ACM SIGCOMM workshop on Network-I/O convergence: experience, lessons, implications
Fast and Robust TCP Session Lookup by Digest Hash

ICPADS '06 Proceedings of the 12th International Conference on Parallel and Distributed Systems - Volume 1
Measuring and Profiling IP Traffic

ECUMN '07 Proceedings of the Fourth European Conference on Universal Multiservice Networks
Algorithms and architectures for network search processors

Algorithms and architectures for network search processors

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper we review local caching of TCP/IP flow context data in the link buffer or a comparable other local buffer. Such connection cache is supposed to be a straight-forward optimization strategy for look-ups of flow context data in a network processor environment. The connection cache can extend common table-based look-up schemes and also be implemented in SW. On the basis of simulations with different IP network traces, we show a significant decrease of average search times. Finally, well-suited cache and table sizes are determined, which can be used for a wide range of IP network systems.