AMON is a software tool for processing multi-10Gbps streams of network data. It is based on PF-RING (zero-copy). This version of AMON reads packets directly from the network interface (NIC), using the PF_RING API. It then generates and streams (to a centralized MongoDB database) the following data output (for more details, see [1]):
- A 128x128 matrix (aka "databrick") that conveys information about the network's traffic intensity and structure
- A 128x128 matrix that conveys information about the network heavy-hitters. These hitters are identified with the help of the MJRTY Boyer-Moore algorithm [2].
The following packages need to be installed:
- libbson-1.0
- libbson-dev
- numactl
- libnuma-dev
- libmongoc-1.0-0
- libmongoc-dev
AMON is currenly supported on Ubuntu and CentoOS systems. So, the first step, is to install one of these on the server that will be receiving the traffic :)
Overall, AMON can be easily built from source, once these two prerequisites are installed:
- PF_RING: we recommend installing PF_RING as a binary. Great instructions can be found here:
http://packages.ntop.org (NOTE: if you want to build PF_RING from source, and then compile AMON, please drop as a line (see "Contact Us" section below).) - MongoDB C drivers: see README.mongo
- Populate accordingly the fields in amon.config
- Type './amon -i eth0' where 'eth0' is the interface you are receiving network traffic from
- Populate the "strata.txt" file (optional). This is for reserving specific "bins" (e.g, the first k=5 leftmost bins) for subnets of interest (e.g., Google, Apple, etc.)
To verify that things work, check that you get updated traffic statistics (provided by PF_RING) every 1 second, and that you get a list of the top-hitters every ALARM_SLEEP seconds (parameter set in amon.config -- we recommend setting this to values less than or equal to 10 seconds)
Please email [email protected] for support or contact mgkallit AT umich
[1] M. Kallitsis, S. Stoev, S. Bhattacharya, G. Michailidis, AMON: An Open Source Architecture for Online Monitoring, Statistical Analysis and Forensics of Multi-gigabit Streams, IEEE JSAC Special Issue on Measuring and Troubleshooting the Internet, July 2016. [Online] http://ieeexplore.ieee.org/document/7460178/
[2] A Linear Time Majority Vote Algorithm. [Online] https://www.cs.utexas.edu/~moore/best-ideas/mjrty/