Wire-Speed Hardware-Assisted Traffic Filtering with Mainstream Network Adapters - Advances in Network-Embedded Management and Applications - page 84

without requiring external switching equipment. From the hardware point of
view, we envisage that future NICs will further enhance flow affinity filters
number and expressiveness (e.g. adding the ability to filter tunneled traffic),
add per-filter statistics (e.g. number of packets and bytes that matched each
filter) so that developers could implement efficient NetFlow caches in
hardware.

7 Conclusions

Monitoring the Internet is challenging as high-speed networks are becoming
popular and traffic patterns more complex. In order to satisfy the increasing
performance requirements and reduce deployment costs, modern network
monitoring frameworks should leverage those features offered by mainstream
NICs that are introduced for general-purpose networking and not fully
exploited in the context of network monitoring. This paper has presented an
evolution of PF_RING, a monitoring framework originally designed for
accelerating packet capture, that exploits hardware-based filtering
mechanisms offered by the Intel 82599 based NICs and likely future NICs.
Thanks to flow affinity filters PF_RING can now fine-grain flow balance
packets across cores, classify traffic and discard unwanted communication
patterns directly into the NIC before packets hit the driver. The validation
process has demonstrated that many network applications can benefit from
this work, making it very general and usable also outside of the network
monitoring domain. Not to mention that it is finally possible to combine the
speed of hardware with the flexibility of software for effectively monitoring
10 Gbit networks using commodity network adapters.

Availability. This work is distributed under the GNU GPL license and is
available at no cost form the PF_RING home page

( http://www.ntop.org/PF_RING.html ).

Acknowledgments. The authors would like to thank Intel and in particular
Edward Clinton and Richard P. Kelly for their support during this research
work.

References

1. W. John and others, Passive internet measurement: Overview and

guidelines based on experiences, Computer Communications, vol. 33,
issue 5 (2010).