saphalpdyl

Working towards building a network fault attribution system oriented towards NaaS platforms that manage their own hardware. In the experimentation phase with existing monitoring frameworks, anomaly detection, and fault localization for sub-optimal routes.

Possibly see a use case in home labs too, but not there yet.

Thank you for the context. I did start out with mininet, then moved to containernet->containerlab. Mininet could not model subscriber session lifecycle in how I wanted it. P4/Openflow is on the radar, thanks for the pointer.

That is awesome! I was walking almost blind folded here, making decisions based on observations, intuition, some blogs, and RFC. These projects give me something to look against as I further develop my skills.

Thank you. I'm exploring more on this. Getting feedback from people here has hyped me up even more for the next step.

I appreciate the compliment. I wish this type of knowledge was more easily available for the general public since it represents an integral part of modern day internet. A comment on this thread mentioned other similar ongoing project which I'm very happy about and excited to explore.

Thank you! That answers my question. I appreciate your feedback.

My access edge is injecting DHCP Option 82, and I'm mapping customers based on (bng_id + circuit_id + remote_id ). Say, a customer on Oakwood Drive ABC wants a service. What is the process of finding the right circuit between storing the customer's desired address and finding the best circuit to connect it to? Since, as mentioned in this thread, having connected to a wrong circuit can cause network noise for other customers too, how is the "cleanest" circuit + port assigned to a customer in a location ?

Subscribed! I am very new to things here since it's barely been a 1.5 months since I touched this area, so many of my questions have probably been answered. I will search around and post a question if can't find an answer.

I plan to take all the feedback I can this week, and work on them on spring break.

Thank you for the comment.

Mininet did help me a lot during the initial phases. The main reason I made the switch to containernet(mininet-fork with containers ) and then to containerlab was because I wanted to run an actual NOS image as part of my topology. That was really what pushed me to try and switch to other options.

Yup, its a different experience. Sometimes, you end up learning something you never even intended to.

As for the circuit planning, I'd guess that they have circuits map on something like Netbox and using a intermediary system that maps a customer's location to the nearest circuit. Though, I don't know how they handle the optimization side of it to prevent cases like what happened to your neighbour.

Hello. Containerlab gives me the virtual network topology ( links through veth pairs, containers etc.). The actual BNG's Control plane ( authentication, authorization, session handling, traffic shaping, events streaming etc. ) is written by me. So it's less running off-the-shelf software running on virtualized hardware, and more writing the software and running it on a virtualized hardware.

At some point, I did use Nokia SR Linux as my access node + relay, but had issues with configuration and Option 82. Later, I wrote one myself.

I recently found out about Apache Netbox that would act as the authoritative source of truth for the network topology and replace majority of aether.config.yaml. In Splynx, I did not see any mention of an external solution. It seems they have their own stack for that.

A better and UX-friendly implementation would have been Netbox + aether.config.yaml -> configuration pipeline -> topology.yaml + <other generated files>.