Published on January 12th, 2016 📆 | 4705 Views ⚑0
Joy — Capturing and Analyzing Network Flow Data
A package for capturing and analyzing network flow data and intraflow data, for network research, forensics, and security monitoring.
Joy is a BSD-licensed libpcap-based software package for extracting data features from live network traffic or packet capture (pcap) files, using a flow-oriented model similar to that of IPFIX or Netflow, and then representing these data features in JSON. It also contains analysis tools that can be applied to these data files. Joy can be used to explore data at scale, especially security and threat-relevant data.
JSON is used in order to make the output easily consumable by data analysis tools. While the JSON output files are somewhat verbose, they are reasonably small, and they respond well to compression.
Joy can be configured to obtain intraflow data, that is, data and information about events that occur within a network flow, including:
- the sequence of lengths and arrival times of IP packets,up to some configurable number of packets,
- the empirical probability distribution of the bytes within the data portion of a flow, and the entropy derived from that value,
- the sequence of lengths and arrival times of TLS records,
- other non-encrypted TLS data, such as the list of offered ciphersuites, the selected ciphersuite, and the length of the clientKeyExchange field,
- the name of the process associated with the flow, for flows originate or terminate on the host on which pcap is running.
Joy is intended for use in security research, forensics, and for the monitoring of (small scale) networks to detect vulnerabilities, threats and other unauthorized or unwanted behavior. Researchers, administrators, penetration testers, and security operations teams can put this information to good use, for the protection of the networks being monitored, and in the case of vulnerabilities, for the benefit of the broader community through improved defensive posture. As with any network monitoring tool, Joy could potentially be misused; do not use it on any network of which you are not the owner or the administrator.
Flow, in positive psychology, is a state in which a person performing an activity is fully immersed in a feeling of energized focus, deep involvement, and joy. This second meaning inspired the choice of name for this software package.
Joy has been successfully run and tested on Linux (Debian, Ubuntu, and CentOS) and Mac OSX. The system has been built with gcc and GNU make, but it should work with other development environments as well.
First, obtain the package from github, and change to the joy directory.To build the package, run “make” in the main directoroy:
This will cause the programs to be compiled, linked, stripped, and copied into the main directory as appropriate. It will also run a test script and a unit test program.
The main program for extracting data features from pcap files or live packet captures is the program pcap2flow, which occupies the src/ subdirectory. It is copied into the main joy directory after a successful build. It can be run from that directory, or installed so that it will automatically run as a daemon on Linux or Mac OSX.
Running and Configuration
To understand how pcap2flow is configured, read one of the configuration files (linux.cfg or macosx.cfg). To process a pcap file in offline mode, run:
[joy]$ ./pcap2flow [ OPTIONS ] filename [ filename2 ... ]
[joy]$ ./pcap2flow bidir=1 output=data.json filename
To run the packet capture in online mode, use the same command form, but have OPTIONS include an interface= command, and omit the filename(s) from the command line. For instance,
[joy]$ sudo ./pcap2flow interface=eth0 bidir=1 output=data.json
There are many command line options, so instead of typing them all onto the command line, you may want to have the program read a configuration file. Two such files come with the distribution, linux.cfg and macosx.cfg. If you want to change the program defaults (and you probably do, in order to capture exactly the data of interest to you), then make a copy of the configuration file. By making a local copy that has a different name, your configuration will not be clobbered if you update the joy package.
To install the package on your system, you will need to first build it. Run the script install-sh (as root, or using sudo) to install the package.
[joy]$ sudo ./install-sh
If you run the script with no arguments, then the default configuration will be installed into the /etc/ directory. To have a different configuration file installed, then use the -c option to the install script:
[joy]$ sudo ./install-sh -c local-config-file.cfg
You can also configure anonymization of addresses, which requires a file containing the internal subnets. The default file for those subnets is internal.net; you can change the configuration with the -a option. Similarly, you can change the watchfile of IP addresses (using the -w option) or the SSH private key used to have files uploaded via scp (using the -k option). To see the full option description for the installer, run that program with the -h option to see the help or “usage” message.
A man page will be built and installed automatically as part of the package. See the file pcap2flow.1, or after the install-sh script has been run, access the man page through “man pcap2flow”.