32 comments

[ 14.0 ms ] story [ 102 ms ] thread
I wonder if https://fastnetmon.com/ and DOTS [1] are roughly the same thing?

[1] https://datatracker.ietf.org/wg/dots/about/

FastNetMon does not implement DOTS yet. Instead, we support API for scrubbing centres/boxes directly.
Also, we sent a number of ideas/suggestions to DOTS. And I do not think that our feedback was taken into consideration.
You sent a slide deck that proposed throwing out all the DOTS work after that work had been underway for more than two years.

https://mailarchive.ietf.org/arch/search/?email_list=dots&q=...

This message was prepared about 14 months ago and shared privately with WG members. So, at least few persons from this WG saw it.
You sat on the deck for a year?
Probably, we went away from the topic. Feel free to raise PR if you want to have DOTS in FastNetMon. We'll review it carefully and if it fits our design guidance we will accept it.

Thanks :)

I'm genuinely interested in differences between DOTS and Fastnetmon. Can you elaborate?

I presume DOTS is transport, while Fastnetmon is detection, correct?

Are there some good open source scrubbing center projects you might be able to recommend?

Also might you have any resources or links to how these scrubbing services are implemented, what heuristics they use etc?

I understand the front end of DDOS mitigation i.e netflows, BGP communities and RTBH, and GRE tunnels to the scrubbing centers. However the details of how the scrubbing centers works is something of a mystery to me.

When looking at any of the big DDOS provider's literature, the scrubbing centers are mostly just opaque boxes with little documentation on how they actually work.

You could consider this thing https://github.com/luigirizzo/netmap-ipfw
Is there any documentation how to set this up fast and reliably on an existing Debian server? The only thing I could find was this: https://fastnetmon.com/wp-content/uploads/2017/03/FastNetMon...
https://github.com/pavel-odintsov/fastnetmon#readme

"What can we do? We can detect hosts in our networks sending or receiving large volumes of packets/bytes/flows per second. We can call an external script to notify you, switch off a server, or blackhole the client.

Why did we write this? Because we can't find any software for solving this problem in the open source world!

What is a "flow" in FastNetMon terms? It's one or multiple UDP, TCP, or ICMP connections with unique src IP, dst IP, src port, dst port, and protocol."

That’s a flow in general, and isn’t specific to FastNetMon.
A flow is defined as a unidirectional sequence of packets with some common properties that pass through a network device. [...] for example, flow records include details such as IP addresses, packet and byte counts, timestamps, Type of Service (ToS), application ports, input and output interfaces, etc.

From here: https://tools.ietf.org/html/rfc3954

And if you're on a FreeBSD box, just pkg install fastnetmon

Glad to see this software exists. Had to help build a poor man's version of it at a previous job and it was half baked due to lack of time.

Hi. I have a doubt regarding the license[0].

Aren't GPLv2 and Apache v2 licenses incompatible. How can they co-exist in the same project? (The copyright file says GPLv2, or is it GPLv2+?)

[0] http://metadata.ftp-master.debian.org/changelogs/main/f/fast...

I think that we do not have Apache licensed code in project, it's just protocol description for protobuf: https://github.com/pavel-odintsov/fastnetmon/blob/2005b4e94e...

And it's not used by version available in Debian repositories at all (just some experiments).

The project itself licensed strictly in terms of GPLv2 (not GPLv2+).

Btw, we have the channel at Freenode! Join us: #fastnetmon at irc.freenode.net! :)
What's generally the state of the art in open source home/small office network monitoring? I would like to know and query/audit communication patterns of my devices. While maintaining privacy -> no cloud based commercial products.