Eliminating receive livelock in an interrupt-driven kernel
ACM Transactions on Computer Systems (TOCS)
An Architecture for High Performance Network Analysis
ISCC '01 Proceedings of the Sixth IEEE Symposium on Computers and Communications
Snort - Lightweight Intrusion Detection for Networks
LISA '99 Proceedings of the 13th USENIX conference on System administration
High-Speed Dynamic Packet Filtering
Journal of Network and Systems Management
A scalable multi-core aware software architecture for high-performance network monitoring
Proceedings of the 2nd international conference on Security of information and networks
High speed network traffic analysis with commodity multi-core systems
IMC '10 Proceedings of the 10th ACM SIGCOMM conference on Internet measurement
Enabling high-performance internet-wide measurements on windows
PAM'10 Proceedings of the 11th international conference on Passive and active measurement
vPF_RING: towards wire-speed network monitoring using virtual machines
Proceedings of the 2011 ACM SIGCOMM conference on Internet measurement conference
A high-performance and scalable multi-core aware software solution for network monitoring
The Journal of Supercomputing
| Hi-index | 0.00 |
The capacity of today's network links, along with the heterogeneity of their traffic, is rapidly growing, more than the workstation's processing power. This makes the task of measuring traffic more problematic every day, especially when off-the-shelf hardware is used. A general solution adopted by the computer industry to achieve better performance is to partition the processing among different computing units, exploiting the implicit or explicit parallelism available on today workstations. Parallelism is in fact growing in two dimensions: physical and logical CPUs (e.g. HyperThreading). Unfortunately, most network measurement systems are engineered to process data in a set of sequential tasks; thus, completely ignoring any form of parallelism provided by the hardware. This paper introduces a new approach to build high performance and scalable network measurement tools. It discusses the problem of dispatching packets to different processing entities and describes a technology able to distribute the flow of incoming packets among different processors in an effective and configurable manner, that avoids any copy and optimizes resource usage.