Nfstream: a network data analysis framework using Python

This would be really cool combined with the new VPC traffic mirroring on AWS.

https://aws.amazon.com/blogs/aws/new-vpc-traffic-mirroring/

Thanks for sharing.

+1 and also I like the readme including live demo very much.

This is pretty much a light Python wrapper around ndpi [1] a C library that contains all the dissectors and protocol support. Probably a very bad idea to use a C library with hand-written parsers for that.

[1] https://www.ntop.org/products/deep-packet-inspection/ndpi/

nDPI is used for traffic classification and metadata extraction as it's the currently most reliable state of the art deep packet inspection library. The "Light Python wrapper" is provided as part of nDPI (python repository under nDPI repo).

nfstream is a framework to compute/aggregate network data in an easy manner using Python. The main goal is to shorten the path from networking space to data science space when building for example ML based traffic classification or anomalies detection.